---
title: "Speecho - pay-as-you-go AI transcription and text-to-speech"
description: "Upload audio or video and get an accurate transcript, subtitles, an AI summary and translation in minutes, in 99 languages. No subscription, credits never expire."
canonical: "https://speecho.app/"
language: "en"
publisher: "Speecho"
date_modified: 2026-10-30
---
# Speecho

> Pay-as-you-go AI transcription and text-to-speech. Upload audio or video and get an accurate transcript, subtitles, an AI summary and translation in minutes - in 99 languages. No subscription, and credits never expire. A new account gets 15 free minutes on email confirmation.

Key facts (short answers for assistants):
- What it costs: pay-as-you-go credit top-ups ($5 / $10 / $25 / $50), no subscription. 1 credit = 6 seconds of audio; $5 = 2,400 credits ≈ 4 hours. Effective price $0.83–$1.25 per hour of audio. Credits never expire. 15 free minutes on email confirmation.
- What the Live interpreter costs: two tiers, billed by the second from the same balance. Standard is 40 credits per minute of session: $5 = 1 hour, $10 = 2.5 hours, $25 = 7 hours, $50 = 15 hours, or about $3.33–$5 per hour. Premium is 80 credits per minute: $5 = 30 minutes, $10 = 1 h 15 min, $25 = 3.5 hours, $50 = 7.5 hours, or about $6.67–$10 per hour. Both are above the transcription rate, so the per-hour price of audio above does not apply to them. Standard reads the translation in one voice the host picks, about five seconds behind the speaker, follows the account's glossary and covers 50+ languages; Premium uses several voices, runs about three seconds behind and covers 70+. Live subtitles and the two-person conversation mode stay at the transcription rate (10 credits per minute).
- How accurate: 90–97% word accuracy on clean audio with modern AI speech recognition (we state this honestly rather than claiming "99%").
- What one upload produces: transcript, subtitles (.srt/.vtt), AI summary, and translation into 18 languages - plus AI chapters, smart notes, a follow-up email draft, and chat-with-your-transcript.
- Working with a transcript: the text can be edited in place; a glossary of names, brands and terms can be given before transcribing; notes come in six styles (general, meeting or sales call, interview, lecture, podcast show notes, and a clean text with the filler words removed); it downloads as plain text, text with timestamps, Markdown, SRT, VTT, Word (.docx) or PDF; it can be shared by a read-only public link that can be switched off again. The first two notes styles and the first two translation languages of a transcription are free, further ones cost a few credits.
- Keep audio (opt-in, off by default): the recording of an upload or a live session is stored next to its transcript, private to the account, for up to 90 days; it plays from any timestamp, downloads as MP3 and can be deleted on its own while the text stays.
- Merging: up to 20 transcripts from the history (uploads, live sessions, or conversations in the same language pair) become one, free of charge, in the order the user sets; timestamps run on, "Speaker N" labels are numbered across the parts, the originals stay, and the audio is joined too when every part kept its audio.
- Text-to-speech: 512 stock voices in three quality tiers (185 Standard, 327 ElevenLabs), billed from the same credit balance (50/110/160 credits per 1,000 characters ≈ 1 minute of speech). Every voice can be previewed free. Dialogues too: up to 8 characters, each with its own voice, joined into one MP3 at the same price per character. Clips in Enhanced and Premium voices also download with SRT or VTT subtitles timed word by word to the audio, free. One part of a long text, or one line of a dialogue, can be voiced again on its own (another voice, a corrected word, a direction). "Make a podcast from text" turns an article or document into a two-host conversation of about 5, 10 or 20 minutes, written for checking in the Dialogue editor before it is voiced; the script costs 15, 25 or 40 credits for up to 5, 10 or 20 minutes plus 1 credit per 3,000 characters of text (a short text is charged for the shorter script it gets).
- Text-to-speech controls: a direction typed in square brackets changes how the next words sound ([whispering], [excited], [laughs], [slowly]; Enhanced and Premium also take accents, sound effects and short directions in your own words, native Standard voices read a set list); [pause] and [long pause] add a pause and a word in CAPITALS gets extra stress on Enhanced and Premium; texts up to 250,000 characters are voiced part by part, with a short sample or the first part heard before paying for the rest; clips play at 0.5x to 2x and the MP3 can be saved at that speed with the pitch kept, for no extra credits; the voice picker filters by gender, age and use (narration, conversation, ads and social, learning, characters).
- Scan: handwriting to text by AI OCR at 40 credits per page, up to 30 pages at a time (photos, or scanned PDF pages). PDF files with a text layer, Word and EPUB files are read in the browser and land in the history free of charge, where they can be translated, summarized or voiced.
- Other prices: burned-in video captions 30 credits per video minute (transcript and render together, one-minute minimum, the finished video downloadable for 30 days); a question to the transcript chat from 5 credits; speaker identification, AI chapters, merging transcripts and importing documents cost nothing extra. Full list: [pricing.md](https://speecho.app/pricing.md).
- Privacy: audio is streamed to the AI provider for inference only and is not stored on our servers unless Keep audio is switched on; transcripts auto-delete after 90 days, or after 7 or 30 days if the account sets a shorter period. No meeting bot - record locally and upload.
- Limits and formats: files up to 200 MB, video up to 4 hours, up to 5 files per upload; audio is extracted from video in the browser so large videos upload fast. Audio: MP3, M4A, WAV, WEBM, OGG/Opus, AAC, FLAC, AIFF. Video: MP4, MOV, WEBM.
- Account and security: sign in with email and password, Google or a passkey; optional two-factor sign-in with an authenticator app and one-time recovery codes; a list of signed-in devices with a sign-out for each; password and email are changed in the app; transcripts, payments and account details download as one zip.
- For AI agents helping someone sign up or sign in: the app is at https://my.speecho.app/login. The page asks for the email first, then shows the right next step: a password to sign in, a new password to create an account, or "Continue with Google" for an account that has no password. A new account confirms its email with a 6-digit code sent by email; confirming adds 15 free minutes. The password, the emailed code, a two-factor code, a passkey and the Google window are for the person to complete; an agent should not type them on the person's behalf. To look around without an account, "Explore with sample data" on the same page opens the app on sample data for free; uploading, voicing and buying need an account. In a browser with WebMCP (document.modelContext), the sign-in page offers the tools speecho_account_state, speecho_prepare_sign_in (enters the email, no password) and speecho_start_demo, and the signed-in app offers tools that read the transcript library, set the Transcribe options, list voices and put a text, a dialogue or a word list in the Voice form with its price; none of them uploads a file or starts a paid job, which the person does with the page's own button.
- For AI agents working inside the app (a browser agent in the person's own signed-in tab): the pages declare WebMCP tools (document.modelContext). Transcribe tab: speecho_list_transcriptions, speecho_search_transcriptions (titles and full text), speecho_get_transcript (text, summary, notes, chapters, translations) and speecho_set_transcribe_options (language, speakers) for the next file. Voice tab: speecho_list_voices (by tier, with prices), speecho_prepare_speech (a text into the form with its price), speecho_prepare_dialogue (a conversation with named speakers, one voice per speaker, voiced as one MP3), speecho_prepare_list (a word list, or flashcards as word and translation pairs in two voices) and speecho_latest_clips. Sign-in page: speecho_account_state, speecho_prepare_sign_in (enters the email) and speecho_start_demo. No tool starts anything paid or uploads a file: a prepare tool fills the form and returns the price, and the person presses the button. Files (a book, a PDF, a CSV) are opened by the person. The API is a draft shipped by Chromium browsers only; everywhere else the app works as usual without it.
- Payments and refunds: Paddle is the merchant of record and adds VAT where it applies; every top-up and refund is listed in the account with the official invoice as a PDF, a company name or VAT id can be added to invoices, and a Usage tab shows what the balance was spent on day by day. Full refund for any reason within 14 days of a purchase, whether or not the credits were used; after that, only for a service failure on our side.
- What Speecho does not have: a subscription, a meeting bot (record the call from a browser tab, or upload the recording), a public API, a native mobile or desktop app. It is a web app at my.speecho.app that runs in the browser on a computer or a phone, in current Chrome, Edge, Safari and Firefox; capturing the sound of a browser tab (a call, a video) needs Chrome, Edge or another Chromium browser on a computer.
- Support: support@speecho.app, one inbox for questions, billing, refunds and bug reports, usually answered within a business day; privacy requests go to privacy@speecho.app.
- For whom: individuals and small teams (journalists, students, researchers, podcasters, YouTubers, lawyers, coaches, sales) who transcribe occasionally and don't want a monthly subscription.
- Languages: the feature, use-case and comparison pages, the free tools and the blog are written in English, Spanish, German, French, Portuguese, Japanese, Italian and Dutch; seven more locales (Polish, Vietnamese, Turkish, Ukrainian, Arabic, Korean, Simplified Chinese) have the homepage and the speech-to-text, text-to-speech, conversation translator, live interpreter and live subtitles pages. Pricing, the legal pages, the changelog, the voice library and the language-pair pages are in English. The app UI is available in 15 languages, Simplified Chinese among them.

## Core pages
- [Home / transcribe audio to text](https://speecho.app/): what Speecho does and how it works.
- [Pricing](https://speecho.app/pricing/): pay-as-you-go tiers, per-hour cost, per-feature credit costs, FAQ.
- [Pricing (plain text)](https://speecho.app/pricing.md): machine-readable pricing summary.
- Every page in Markdown: append `index.md` to the URL of any page on this site (`https://speecho.app/live-interpreter/index.md`), or request the page itself with `Accept: text/markdown`. The same text as the page, without the navigation, scripts and players; localized pages come in their own language.
- [Speech to text](https://speecho.app/speech-to-text/): the three routes in - live subtitles while you speak, an uploaded recording, or recording in the browser - plus accuracy, languages and how dictation differs from transcription. Also at /{locale}/speech-to-text/ in de, es, fr, it, nl, pl, pt, vi, tr, uk, ar, ja, ko.
- [Text to speech](https://speecho.app/text-to-speech/): voices, tiers and TTS pricing.
- [Multi-voice text to speech](https://speecho.app/multi-voice-text-to-speech/): dialogues with a voice per character. Paste a script with names ("Anna: ...") and every name becomes a character with its own voice from the same catalogue; the lines are voiced in order with the chosen pause between them (none, 0.3, 0.6 or 1 second) and joined into one MP3. Up to 8 characters and 100 lines, characters can speak different languages, voice directions in square brackets work per line, and editing a voiced dialogue re-voices only the changed lines; one line can also be voiced again on its own. Subtitles for the whole dialogue (SRT/VTT, with or without the speakers' names) on Enhanced and Premium. Priced exactly like the same text in one voice, one tier per dialogue. Also in es, de, fr, pt, ja, it, nl.
- [Conversation translator](https://speecho.app/conversation-translator/): two people, two languages, one phone between them. Each person reads the other in their own language on their half of the screen (the top half is turned to face across the table; on a wide screen the halves sit side by side), live, in any two of 60 languages. Hands-free mode tells who is speaking by the language; hold-to-talk gives each side a button for noisy rooms and close languages (Spanish/Portuguese, Ukrainian/Russian). A glossary of names and terms, with a line about the topic, keeps spellings right. The conversation is saved in the history in both languages and can be exported, summarized, continued or merged with others of the same pair. The translation is text by default; with "Speak the translation aloud" on, each turn is also read to the other person in their language, in a voice picked for each of the two (the microphone pauses while it plays, unless headphones are used). One device, needs internet. Billed like live subtitles: 10 credits a minute, by the second; 40 credits a minute with the translation read aloud. Written in 14 languages at /{locale}/conversation-translator/.
- [Live interpreter](https://speecho.app/live-interpreter/): real-time voice translation - speak, or play a call or a video in a browser tab, and hear it translated in a natural voice a few seconds later, with the text on screen. Two tiers, chosen before each session: Premium (several voices, about three seconds behind the speaker, 70+ languages) and Standard (one voice the host picks, about five seconds behind, the account's glossary of names and terms, 50+ languages). A QR listen link lets up to 20 people listen on their own phones with no app. With an account the link is permanent: it belongs to the account, is the same on every device and for every later session (so it can be printed once as a sign), exists before a session starts, so waiting is free, and can be replaced with a new one at any time. One target language per session. Speech that is already in the target language reaches the translation as text and, with a switch, is voiced too. The original and the translation are both saved to the history, and Keep audio can keep a recording of the original. Translating a browser tab needs Chrome or Edge on a computer. Standard 40 and Premium 80 credits per minute, billed by the second from the same pay-as-you-go balance ($5 = 1 hour on Standard, 30 minutes on Premium). Written in 14 languages at /{locale}/live-interpreter/.
- Text-to-speech pages written in the reader's own language, for people searching in it rather than in English: [Text in Sprache umwandeln](https://speecho.app/de/text-to-speech/german/), [Озвучити текст українською](https://speecho.app/uk/text-to-speech/ukrainian/), [تحويل النص إلى كلام بالعربية](https://speecho.app/ar/text-to-speech/arabic/).
- English accent pages, one per accent, each with playable samples and the accent differences spelled out: [American](https://speecho.app/text-to-speech/american-english/), [British](https://speecho.app/text-to-speech/british-english/), [Australian](https://speecho.app/text-to-speech/australian-english/), [Indian](https://speecho.app/text-to-speech/indian-english/), [Irish](https://speecho.app/text-to-speech/irish-english/), [Scottish](https://speecho.app/text-to-speech/scottish-english/), [Canadian](https://speecho.app/text-to-speech/canadian-english/), [New Zealand](https://speecho.app/text-to-speech/new-zealand-english/), [South African](https://speecho.app/text-to-speech/south-african-english/).
- Regional variants of other languages, one page each, with voices from that country, the sound differences spelled out and the local vocabulary a script needs: [Mexican Spanish](https://speecho.app/text-to-speech/mexican-spanish/) (7 voices), [Castilian Spanish, Spain](https://speecho.app/text-to-speech/castilian-spanish/) (7 voices), [Canadian French, Québec](https://speecho.app/text-to-speech/canadian-french/) (6 voices), [European Portuguese, Portugal](https://speecho.app/text-to-speech/european-portuguese/) (6 voices), [Flemish, Belgian Dutch](https://speecho.app/text-to-speech/flemish/) (6 voices), [Saudi Arabic](https://speecho.app/text-to-speech/saudi-arabic/) (8 voices: Najdi, Hijazi, Eastern Province), [Gulf Arabic, Khaleeji](https://speecho.app/text-to-speech/gulf-arabic/) (8 voices: Emirati, Kuwaiti). In the app these voices carry the tags MX, ES, Québec, PT, BE, Saudi and Gulf next to their names.
- English spoken WITH a foreign accent - a different thing from the native-language pages, for characters, dubbing and language teaching: [German accent](https://speecho.app/text-to-speech/german-accent/), [French accent](https://speecho.app/text-to-speech/french-accent/), [Italian accent](https://speecho.app/text-to-speech/italian-accent/), [Japanese accent](https://speecho.app/text-to-speech/japanese-accent/). These read English text; for German or Japanese text use the native-language pages instead.
- [TTS voice library](https://speecho.app/text-to-speech/voices/): all 512 voices with playable samples - native speakers of 31 languages on Enhanced and Premium, 185 native voices in 19 languages on Standard, 6 universal ElevenLabs voices.
- Per-language TTS pages with native-voice samples, four to twenty-five voices each (Arabic 25: Modern Standard, Saudi, Gulf, Egyptian, Levantine, Moroccan; Spanish 21: Latin American, Mexican, Castilian; French 15: France, Québec; Portuguese 15: Brazil, Portugal; Dutch 11: Netherlands, Flemish): [English](https://speecho.app/text-to-speech/english/), [Spanish](https://speecho.app/text-to-speech/spanish/), [German](https://speecho.app/text-to-speech/german/), [French](https://speecho.app/text-to-speech/french/), [Portuguese](https://speecho.app/text-to-speech/portuguese/), [Italian](https://speecho.app/text-to-speech/italian/), [Japanese](https://speecho.app/text-to-speech/japanese/), [Arabic](https://speecho.app/text-to-speech/arabic/), [Korean](https://speecho.app/text-to-speech/korean/), [Dutch](https://speecho.app/text-to-speech/dutch/), [Polish](https://speecho.app/text-to-speech/polish/), [Turkish](https://speecho.app/text-to-speech/turkish/), [Ukrainian](https://speecho.app/text-to-speech/ukrainian/), [Vietnamese](https://speecho.app/text-to-speech/vietnamese/), [Hindi](https://speecho.app/text-to-speech/hindi/), [Indonesian](https://speecho.app/text-to-speech/indonesian/), [Chinese](https://speecho.app/text-to-speech/chinese/), [Russian](https://speecho.app/text-to-speech/russian/), [Filipino](https://speecho.app/text-to-speech/filipino/), [Malay](https://speecho.app/text-to-speech/malay/), [Swedish](https://speecho.app/text-to-speech/swedish/), [Romanian](https://speecho.app/text-to-speech/romanian/), [Czech](https://speecho.app/text-to-speech/czech/), [Greek](https://speecho.app/text-to-speech/greek/), [Hungarian](https://speecho.app/text-to-speech/hungarian/), [Norwegian](https://speecho.app/text-to-speech/norwegian/), [Danish](https://speecho.app/text-to-speech/danish/), [Finnish](https://speecho.app/text-to-speech/finnish/), [Bulgarian](https://speecho.app/text-to-speech/bulgarian/), [Croatian](https://speecho.app/text-to-speech/croatian/), [Slovak](https://speecho.app/text-to-speech/slovak/).
- Text-to-speech in the reader's own language at /{locale}/text-to-speech/ - de, es, fr, it, nl, pl, pt, vi, tr, uk, ar, ja, ko, zh - each with playable samples from that language's native speakers. Six of them (pl, vi, tr, uk, ar, ko) also carry a native-versus-universal voice comparison and pronunciation questions answered in that language.
- The Czech, Greek, Hungarian, Norwegian, Danish, Finnish, Bulgarian, Croatian and Slovak pages also exist in that language itself, each the only page of the site in it (the app is in English): [čeština](https://speecho.app/cs/text-to-speech/czech/), [Ελληνικά](https://speecho.app/el/text-to-speech/greek/), [magyar](https://speecho.app/hu/text-to-speech/hungarian/), [norsk](https://speecho.app/no/text-to-speech/norwegian/), [dansk](https://speecho.app/da/text-to-speech/danish/), [suomi](https://speecho.app/fi/text-to-speech/finnish/), [български](https://speecho.app/bg/text-to-speech/bulgarian/), [hrvatski](https://speecho.app/hr/text-to-speech/croatian/), [slovenčina](https://speecho.app/sk/text-to-speech/slovak/).
- The regional pages also exist in their own language, under the site's locales: [Mexican Spanish in Spanish](https://speecho.app/es/text-to-speech/mexican-spanish/), [Castilian Spanish in Spanish](https://speecho.app/es/text-to-speech/castilian-spanish/), [Québec French in French](https://speecho.app/fr/text-to-speech/canadian-french/), [European Portuguese in Portuguese](https://speecho.app/pt/text-to-speech/european-portuguese/), [Flemish in Dutch](https://speecho.app/nl/text-to-speech/flemish/), [Saudi Arabic in Arabic](https://speecho.app/ar/text-to-speech/saudi-arabic/), [Gulf Arabic in Arabic](https://speecho.app/ar/text-to-speech/gulf-arabic/); Swedish and Romanian in their own language, the only page of the site in each: [svenska](https://speecho.app/sv/text-to-speech/swedish/), [română](https://speecho.app/ro/text-to-speech/romanian/).
- [About](https://speecho.app/about/): who is behind Speecho and how we handle audio.
- [Privacy](https://speecho.app/privacy/): data handling and 90-day retention.
- [Security](https://speecho.app/security/): how audio and accounts are protected (TLS on every connection, encryption at rest, audio kept only when asked for) and how to report a vulnerability.
- [Terms](https://speecho.app/terms/): terms of service - credits, refunds, acceptable use, and your rights to what you transcribe and generate.
- [Refund policy](https://speecho.app/refund/): full refund for any reason within 14 days of a purchase; after that, for service failures on our side.
- [Help](https://speecho.app/help/): questions and answers about Speecho as a whole, searchable: credits and prices, refunds, file formats and limits, live sessions, text to speech, privacy and the account. Written in 14 languages at /{locale}/help/.
- [Contact](https://speecho.app/contact/): the support inbox, and what to include so the first reply is the fix.
- [Changelog](https://speecho.app/changelog/): what shipped and when, newest first - the place to check whether a feature exists yet.
- [Changelog (plain text)](https://speecho.app/changelog.md): the same list as markdown.

## Free tools (no signup; they run in the browser, with one stated exception: the teleprompter's optional AI block)
All tools are available in en/es/de/fr/pt/ja/it/nl - prepend the locale to the path (e.g. /de/tools/…).
- [Tools hub](https://speecho.app/tools/)
- [Transcription cost calculator](https://speecho.app/tools/transcription-cost-calculator/): audio length → pay-as-you-go price.
- [Text-to-speech cost calculator](https://speecho.app/tools/tts-cost-calculator/): script → credits, price and speech length per voice tier.
- [Speaking time calculator](https://speecho.app/tools/speaking-time-calculator/): script → speaking time at an adjustable pace.
- [Subtitle converter](https://speecho.app/tools/subtitle-converter/): SRT ↔ VTT ↔ SBV ↔ plain text, plus a transcript mode that labels speakers (Zoom, Teams, "Speaker 1:") and joins cues into paragraphs, and a merge that stacks an original and its translation into one bilingual caption file.
- [Subtitle timing shifter](https://speecho.app/tools/subtitle-shifter/): shift timestamps or rescale for a different framerate.
- [Caption safe zone checker](https://speecho.app/tools/caption-safe-zone-checker/): drop a vertical video or a frame, paste an SRT, and see which captions land in the strip the interface covers. Uses Meta's published ad-guide margins (14% top / 35% bottom / 6% sides, 40% bottom with a disclaimer); has no TikTok preset on purpose, because TikTok publishes no numbers - you overlay a screenshot from your own phone instead. Exports a 1080x1920 overlay PNG.
- [Subtitle readability checker](https://speecho.app/tools/subtitle-readability-checker/): flags captions over CPS/line-length/duration limits, and can repair what it flagged - re-wrapping lines, splitting captions that cannot fit, and stretching short ones into adjacent silence without moving a neighbour - then hands back the fixed file and re-checks its own output.
- [YouTube chapter formatter](https://speecho.app/tools/youtube-chapter-formatter/): rough timestamps → a list that meets YouTube's chapter rules, or Podcasting 2.0 chapters as a JSON file or a podcast:chapters RSS tag.
- [Filler word counter](https://speecho.app/tools/filler-word-counter/): counts um's, uh's and crutch words in 12 languages, with highlights. Takes an SRT or VTT as well as plain text - timecodes fill in the recording length and put the fillers on a per-minute timeline showing where they cluster - and can strip them out, tidying the punctuation left behind and collapsing stutters, returning a subtitle file with the original timings or a clean text file.
- [Voice recorder](https://speecho.app/tools/voice-recorder/): records from the microphone and hands back a downloadable file; markers can be dropped live with the M key or from a list of prepared questions, and export as a timestamped text file. Nothing is uploaded and the recording only exists in the tab.
- [Video recorder](https://speecho.app/tools/video-recorder/): records the camera, the screen (a tab, a window or the whole screen) or the screen with the camera in a round frame in the bottom-right corner, and hands back a downloadable file: MP4 (H.264/AAC) in recent Chrome, Edge and Safari, WebM in Firefox. Microphone on or off; sound from a shared tab can be mixed in where the browser offers it (Chrome and Edge). In camera mode an optional script scrolls over the preview near the lens at a set words-per-minute pace and is not part of the video; there is a 3-second countdown, pause, and markers (M key) that export as a timestamped text file. Up to 30 minutes per take, held in memory; screen modes need a desktop browser. Markers can be clicked afterwards to jump to that moment in the finished video. On phones only the camera mode works (no browser on a phone can capture the screen). Nothing is uploaded.
- [Voice typing](https://speecho.app/tools/voice-typing/): dictation in the browser through the Web Speech API, in 30 languages; editable text, copy, .txt download, English voice commands for punctuation, a draft kept in the browser. Recognition is done by the browser's own speech service (Chrome: Google, Edge: Microsoft, Safari: Apple), not by Speecho; Firefox has no such service and gets a pointer to live subtitles instead. Hands the text to the text-to-speech panel in one click.
- [Conversation captions](https://speecho.app/tools/conversation-captions/): a caption screen for talking face to face, for people who are deaf or hard of hearing and the people they talk to. One phone, tablet or laptop sits on the table and shows what is said in large text, the newest phrase brightest, in 30 languages. Text size in seven steps, white on black, black on white or yellow on black, full screen, a flip that turns the text to face the person sitting opposite, a manual "Next speaker" button that gives the other person a second colour, and a typed reply shown in the same large text. Recognition is done by the browser's own speech service (Chrome: Google, Edge: Microsoft, Safari: Apple), not by Speecho, so it is free with no account; Firefox has no such service. Nothing is saved. It does not translate, tell voices apart, send the text to other screens or keep a transcript: translation between two languages is the conversation translator, and captions with translation and a viewer link are Speecho Live subtitles.
- [Read aloud](https://speecho.app/tools/read-aloud/): reads pasted text or an opened file aloud with the browser's own speech synthesis (Web Speech API): PDF with a text layer, Word .docx, EPUB, TXT, Markdown, SRT/VTT/SBV with the timing stripped. Play, pause, stop, 0.5x to 2x speed, a voice list grouped by language that picks a voice for the detected text language, the current sentence highlighted (and the word, where the browser reports word boundaries), click a sentence to jump to it; copy and .txt download. Files are parsed in the tab (pdf.js, mammoth, JSZip) and never uploaded, up to 200,000 characters; scanned PDF pages have no text layer and are skipped and counted, not OCR'd. Voice quality depends on the device, and voices the browser marks as online (Google voices in Chrome, Online voices in Edge) send each sentence to Google or Microsoft; the page labels them and prefers on-device voices. A browser voice cannot be saved, so there is no MP3: the page hands the text to the Speecho text-to-speech panel for that.
- [Teleprompter](https://speecho.app/tools/teleprompter/): scrolls a script at a pace set in words per minute rather than arbitrary units, with a mirror mode for prompter glass, a vertical flip, pause marks and highlighted words, saved scripts, PDF/Word/TXT import and a QR link that moves a script to a phone. It can follow the reader's voice instead of a fixed pace (browser speech recognition in Chrome, Edge and Safari: it waits through an ad-lib and goes back with a repeated sentence), and a run in that mode ends with a rehearsal report worked out on the device: average pace, the fastest and slowest paragraphs, pauses of three seconds or more with the words that followed, and the places where the reader went back or skipped. An AI block writes a script from a few notes or shortens an existing script to a chosen length (0:30 to 10:00 at the reader's own pace), free up to 5 times a day with no account; a script that already fits the length is not sent at all; this is the only part of the page that sends text to a server (Speecho and an AI provider, not stored), and the result's length is close to the target rather than exact. A "Record a video with this script" button opens the video recorder with the script and its pace already loaded (passed in the URL fragment, not sent to a server).
- [Audio speed changer](https://speecho.app/tools/audio-speed-changer/): plays audio or video at 0.5x to 3x with the pitch kept normal, and shows how much time each speed saves; a player, not a converter.
- [Silence remover](https://speecho.app/tools/silence-remover/): shortens gaps under a threshold and shows every one of them on the waveform first; exports a trimmed WAV.
- [Audio quality checker](https://speecho.app/tools/audio-quality-checker/): drop a recording and get its speech level, background noise, signal-to-noise ratio, clipping, silence share and a dead-channel check, with a good / usable / risky verdict for transcription; a second mode tests a microphone live (a moment of room noise, then one sentence), either raw for a recording or the way a live captions or interpretation session hears it, with noise suppression on, and that result can be sent on as a link that carries only the numbers. The thresholds are a rule of thumb for speech, not a published standard; the live ones rest on a test with synthetic voices and added noise (2026-10-04). Nothing is uploaded.
- [Extract audio from video](https://speecho.app/tools/extract-audio-from-video/): pulls the soundtrack out of an MP4, MOV, MKV or WebM as MP3, in the browser, with nothing uploaded. Before the converter loads it shows the length, the transcription price and the output size; it can keep only a time range, shrink a large audio file too, and splits anything that would exceed the app's 200 MB upload limit into parts.
- [Replace audio in video](https://speecho.app/tools/replace-audio-in-video/): puts a new soundtrack (a voice-over, a translation, music) on an MP4, MOV, WebM or MKV, in the browser with nothing uploaded. The video stream is copied, not re-encoded; only the sound is encoded (AAC, or Opus for WebM). The old sound is removed or kept quietly underneath at a chosen level (the original speech stays in it: a mixed track is not separated). A shorter sound is padded with silence to the end of the video, a longer one cut there; the lengths are compared before anything runs. Up to about 1 GB on a computer, 300 MB on a phone. Also in es, de, fr, pt, ja, it, nl. The last step of the app's voice-over made from a video's own transcript (Text to speech, Subtitles mode), whose track is as long as the video.
- [Voice note converter](https://speecho.app/tools/voice-note-converter/): turns WhatsApp .opus, Telegram .oga, .ogg and old-phone .amr/.3gp voice messages into MP3, up to 20 files at once, in the browser with nothing uploaded. Opus/Ogg are decoded by the browser itself; AMR and 3GP load ffmpeg.wasm once. Notes that Speecho transcribes .opus/.ogg/.oga directly, so converting is only needed for AMR/3GP.
- [Merge audio files](https://speecho.app/tools/merge-audio-files/): joins up to 20 recordings (voice messages, split recorder files, lecture parts) into one MP3, in the browser with nothing uploaded. Files can be reordered or sorted by name, a pause of 0.5 to 2 s can go between them, and the volume is evened out across files; output is Speech (64 kbps mono) or Music (128 kbps stereo). Limits: 2 hours per file, 5 hours in total. Before merging it shows the total length, the MP3 size, the transcription price and whether the result fits the app's 200 MB upload limit. Notes that merging is not needed for a transcript: Speecho takes up to 5 files per upload and merges 2 to 20 transcripts into one.
- [MP3 to M4B](https://speecho.app/tools/mp3-to-m4b/): joins MP3, M4A, WAV, FLAC or OGG files into one M4B audiobook file, one chapter per file (titles from the file names, editable, reorderable), or one long recording cut into chapters by typed times ("4:35 Chapter 1"); adds the title, author, a cover image and the audiobook flag, so Apple Books, BookPlayer, Smart AudioBook Player, VLC and Audiobookshelf show the chapters and remember the position. AAC at 64 kbps mono (speech) or 128 kbps stereo, or M4A input packed without re-encoding. Runs in the browser with ffmpeg.wasm, nothing uploaded; up to 12 hours on a computer, 3 hours on a phone.
- [Remove background noise](https://speecho.app/tools/remove-background-noise/): cleans steady background noise (hum, hiss, fans, traffic, rain, keyboards) out of a voice recording with DeepFilterNet 3, an open-source model that runs in the browser; the recording is never uploaded, only the model (about 12 MB, once) is downloaded from speecho.app. Upload a file (up to 60 minutes) or record on the page (up to 30 minutes, browser noise suppression off), pick Light, Medium or Strong, compare Original and Cleaned in one player, and download MP3 (128 kbps) or WAV, mono 48 kHz. Shows background noise and voice-above-noise before and after. Does not separate a voice from music or other speakers. Notes that cleaning is usually not needed before transcription.
- [Quizlet to Anki converter](https://speecho.app/tools/flashcard-converter/): a Quizlet export, an Anki "Notes in Plain Text" export, a CSV or any typed word list (tab, comma, semicolon, arrow, dash, colon, equals) into an Anki import file with header lines, Quizlet import text, or a clean CSV; removes numbering, duplicates, headings and Anki HTML, can swap sides; runs in the browser.
- [CSV to Anki deck](https://speecho.app/tools/csv-to-anki-deck/): turns a CSV, a Quizlet or Anki export or a pasted word list into an Anki .apkg deck, optionally with sound on each card from MP3s or the ZIP + Anki CSV that Speecho's flashcard audio produces. Runs in the browser, nothing is uploaded.
- [Live translation sign maker](https://speecho.app/tools/live-translation-sign/): turns a Speecho Live interpreter listener link (my.speecho.app/listen/CODE) or a Live subtitles link (/watch/CODE) into a printable scan-to-listen sign: a QR code, the address as text, a heading and three steps (scan, tap Listen, use headphones) in one of 14 languages (the one the session translates into) or in your own text, optional event name and Wi-Fi. Formats: A4 or Letter sheet, folded table tent, 16:9 slide, four cards per sheet; prints (or saves as PDF) and downloads as PNG. Accepts Speecho links only; the link stays in the browser and nothing is uploaded.

## Video walkthroughs
Short screen recordings of the real app on [Speecho's YouTube channel](https://www.youtube.com/@speecho-app), in English with dubbed audio or subtitles in es, de, fr, pt, ja, it and nl. Useful when someone asks to see how a feature works.
- [Transcribe audio or video to text](https://www.youtube.com/watch?v=6VUGfwEwmSo): drop a file, see the length and price, get the transcript with timestamps, a summary and a translation.
- [Text to speech and MP3 download](https://www.youtube.com/watch?v=AAIwuHhr5j0): type or paste text, preview voices for free, download the MP3.
- [Live subtitles and live translation in the browser](https://www.youtube.com/watch?v=Wersl4vSYY0): speak or play a meeting through the microphone and read captions in real time, translated if you want.
- [Face-to-face conversation in two languages](https://www.youtube.com/watch?v=06GEzIJiuE8): one screen between two people, each reading the other in their own language.
- [Live interpreter](https://www.youtube.com/watch?v=2RjUyeel8u0): your speech translated live into another language as a natural voice and as text, listeners on their own phones.
- [Free teleprompter that follows your voice](https://www.youtube.com/watch?v=_Q2EukseBBE): paste a script, set the size and pace, and let it scroll with your voice.
- [AI dialogue with a voice for each character](https://www.youtube.com/watch?v=SMr6w-GqSVM): name the characters, give each a voice, download one MP3.
- [Text to speech with a British, Indian or Australian accent](https://www.youtube.com/watch?v=JBc1DcfPh7w): English voiceovers in the accent the audience expects.
- [Live subtitles in OBS, with live translation](https://www.youtube.com/watch?v=qCLeUC0ca80): copy a link from Speecho into an OBS browser source.

## Blog (English; every post also in es/de/fr/pt/ja/it/nl under /{locale}/blog/)
One-file version of the summary and every post: [llms-full.txt](https://speecho.app/llms-full.txt).
- [Blog index](https://speecho.app/blog/): every post, newest first, also grouped by topic (how-to, guides, meetings, creators, accuracy).
- [How to transcribe audio to text](https://speecho.app/blog/how-to-transcribe-audio-to-text/): the four options compared (manual, free tools, AI, human).
- [What you can cut from a quote](https://speecho.app/blog/what-you-can-cut-from-a-quote/): what AP, the New York Times and the Washington Post style guidance allows when altering a direct quote - hesitations and false starts may go silently, grammar may never be fixed, ellipses mark removed content and not removed noise, and AP dropped (sic) in 2019. Includes the rule that an unverified AI transcript is not a quotable source: quotes get checked against the audio at their timecode.
- [Clean verbatim vs full verbatim](https://speecho.app/blog/clean-verbatim-vs-full-verbatim/): what each level removes (fillers, false starts, stutters, non-speech sounds), who needs which, and the honest position that raw AI output is neither - models trained on captions drop an unpredictable share of hesitations, so an unreviewed AI transcript does not meet a full-verbatim standard.
- [How accurate is AI transcription in 2026?](https://speecho.app/blog/how-accurate-is-ai-transcription/): honest 90–97% answer and what affects it.
- [Is it safe to upload audio to a transcription tool?](https://speecho.app/blog/is-it-safe-to-upload-audio/): privacy explainer.
- [Transcribe a meeting without a bot](https://speecho.app/blog/transcribe-a-meeting-without-a-bot/): no-bot workflow for Zoom/Meet/Teams.
- [Live captions in Zoom without a bot](https://speecho.app/blog/live-captions-in-zoom-without-a-bot/): the caption API token step by step (free Zoom accounts too), with live translation; YouTube Live and OBS covered, Teams on request, Google Meet not possible.
- [Podcast episode to show notes](https://speecho.app/blog/podcast-episode-to-show-notes/): transcript → show notes, chapters and quotes.
- [How to make an AI voiceover](https://speecho.app/blog/how-to-make-an-ai-voiceover/): script → MP3 narration, with subtitles.
- [Turn an article into a podcast](https://speecho.app/blog/turn-an-article-into-a-podcast/): two AI hosts from a text or PDF: checking the script, voices, length, cost.
- [ElevenLabs vs OpenAI TTS](https://speecho.app/blog/elevenlabs-vs-openai-tts/): voice quality and price compared.
- [How to convert a Zoom transcript to Word](https://speecho.app/blog/zoom-transcript-to-word/): .vtt → clean Word doc, free converter or DOCX export.
- [Professional phone greeting scripts](https://speecho.app/blog/professional-phone-greeting-scripts/): 8 voicemail/IVR templates + AI-voice MP3 how-to.
- [Turn any video into text](https://speecho.app/blog/turn-any-video-into-text/): upload the video or extract the audio first, and what changes either way.
- [Interview transcription for journalists](https://speecho.app/blog/interview-transcription-for-journalists/): speaker labels, quotes and the accuracy question.
- [Subtitle timing rules](https://speecho.app/blog/subtitle-timing-rules/): the four broadcast numbers - 17 characters a second, line length, line count, minimum duration.
- [Record a lecture and turn it into study notes](https://speecho.app/blog/lecture-recording-to-study-notes/): the short answer first (record with a phone or the free browser voice recorder, dropping markers with M; upload for a transcript, chapters, a summary and answers to your own questions; live subtitles if the words are needed during the lecture), then recording so the transcript is usable and turning it into notes you can revise from.
- [How to transcribe iPhone Voice Memos](https://speecho.app/blog/transcribe-iphone-voice-memos/): Apple's built-in transcription needs iPhone 12 and iOS 18 but not Apple Intelligence; ten languages, no speaker labels, no file export.
- [How to transcribe WhatsApp voice messages](https://speecho.app/blog/transcribe-whatsapp-voice-messages/): on-device transcripts, four languages on Android against twenty on iPhone, and nothing at all on Web or desktop.
- [How to get the transcript of a YouTube video](https://speecho.app/blog/get-youtube-video-transcript/): the Show transcript panel, downloading captions from Studio, and what automatic captions get wrong.
- [Subtitles for Reels and TikTok](https://speecho.app/blog/add-subtitles-to-reels-and-tiktok/): neither platform publishes a safe zone for organic posts, the mute statistic is from 2016, and readability is the part you control.
- [How to transcribe a Microsoft Teams recording](https://speecho.app/blog/transcribe-teams-meeting-recording/): meeting recordings land in the organizer's OneDrive "Recordings" folder (channel meetings in the channel's SharePoint), attendees can view but by default not download, recordings and transcripts expire after 120 days by default (30 on A1); the Recap-tab transcript (.docx or .vtt, organizers only) versus transcribing the MP4 itself.
- [Handwritten notes to text](https://speecho.app/blog/handwritten-notes-to-text/): six rules for photographing a handwritten page, when a scanner is worth it (300 vs 600 dpi), what AI handwriting recognition still misreads (faded ink, margins, tables, historical scripts like Kurrent), and a three-step check of the result.
- [ADA Title II and your university's video](https://speecho.app/blog/ada-title-ii-video-captions-universities/): who has to caption video under the DOJ web rule (WCAG 2.1 AA), the deadlines as extended by the April 2026 interim final rule (26 April 2027 / 26 April 2028), the five exceptions, the three WCAG criteria that decide video, and what each route costs per hour - including the note that a human proofreading pass is not optional.
- [How to write meeting minutes](https://speecho.app/blog/how-to-write-meeting-minutes/): the format that works (decisions, owners, dates), a template to copy, a full sample from a 30-minute meeting, and the route from a recording to finished minutes.
- [How to add a voiceover to PowerPoint and Google Slides](https://speecho.app/blog/voiceover-for-powerpoint-and-google-slides/): record your own voice or generate it from the script; PowerPoint's Record Audio and Recording tab, Google Slides' audio-from-Drive route and its sharing trap, and exporting a narrated deck as a video.
- [How to write a speech](https://speecho.app/blog/how-to-write-a-speech/): one sentence first, words per minute for a time slot (130 calm, 160 brisk; characters per minute in the Japanese version), opening / three points / close, a two-minute sample adapted per language, a template, and rehearsing with the speaking time calculator, recorders, filler counter and teleprompter.
- [Bad audio? What to fix before transcribing](https://speecho.app/blog/fix-bad-audio-before-transcription/): check the recording first (signal-to-noise 25 dB comfortable, under 15 dB expect gaps), which problems a transcript survives and which no software fixes (clipping, overlapping speech), a free sample before any cleanup, trimming silence to pay for less audio, and a caution on AI enhancers.
- [Where AI voices are allowed in 2026](https://speecho.app/blog/where-ai-voices-are-allowed/): YouTube monetization and disclosure, Audible/ACX (human narration unless authorized), Spotify and Apple Podcasts (allowed with disclosure), Apple Books and Google Play (own narration programs), TikTok and Meta labels, Google and Meta political ads, US outbound calls under the TCPA and the EU AI Act Article 50 from 2 August 2026; each row checked against the platform's own page in October 2026 and linked.

## Comparisons
- [Alternatives hub](https://speecho.app/alternatives/): Speecho vs Otter, Rev, Notta, TurboScribe, Happy Scribe, Sonix.
- One page per competitor, each an honest comparison of price and fit: [Otter.ai](https://speecho.app/otter-alternative/) (pay-as-you-go against monthly plans, no bot in the call), [Rev](https://speecho.app/rev-alternative/), [Notta](https://speecho.app/notta-alternative/), [TurboScribe](https://speecho.app/turboscribe-alternative/), [Happy Scribe](https://speecho.app/happy-scribe-vs-speecho/), [Sonix](https://speecho.app/sonix-vs-speecho/).
- [Whisper online](https://speecho.app/whisper-online/): for people who would otherwise install OpenAI Whisper - the same job in the browser, with no Python and no install.
- [Transcription without a subscription](https://speecho.app/transcription-without-subscription/): why pay-as-you-go beats a monthly plan.

## Popular use cases
- [Record a meeting without a bot](https://speecho.app/record-meeting-without-bot/): record Google Meet, Zoom or Teams from the browser tab (Chrome or Edge on a computer), with no bot joining and nothing to install, and get a transcript with speakers and AI notes. The page says how to open each platform in the browser: Google Meet always runs there, Teams via "Continue on this browser" on the join page, Zoom via the "Join from your browser" link (if the host allows it).
- [Add captions to video](https://speecho.app/add-captions-to-video/): burned-in subtitles for Reels, TikTok, Shorts and course clips; six styles including three word-by-word highlight modes (recolour, pop, fill) driven by the transcript's word timestamps, position, size and colours previewed on a frame of your own video, a Reels safe-zone shortcut, CJK/Thai/Devanagari fonts, pay per minute, no watermark.
- [Live subtitles](https://speecho.app/live-subtitles/): real-time captions and translation in the browser, shareable viewer link, an OBS overlay with a permanent link for the account (pasted into OBS once, it shows every later session without making it public) that is restyled live from the app (colour, alignment, text size, outline or a dark band, fade-out after silence), pause without losing the session, optional private recording that can later be re-transcribed with the full-accuracy engine; a finished session can be continued later, appending to the same transcript, recording and share link; a glossary keeps names and terms right: names and terms (the same list file transcription uses), translations of terms per target language written as term = translation, and a line about the topic, set before or changed during a session
- Live subtitles as the meeting's own captions, with no bot in the call: [live captions for Zoom](https://speecho.app/live-captions-for-zoom/) (the host pastes Zoom's caption API token; free Zoom accounts included). The captions can be the original or a live translation into any of 60 languages; the usual Live rate, 10 credits a minute. YouTube Live (caption ingestion URL) and OBS (captions embedded in the stream) work the same way. Microsoft Teams (CART captions) is not offered yet, on request; Google Meet has no such option. The page also in es, de, fr, pt, ja, it, nl.
- [Live captions for OBS](https://speecho.app/live-captions-for-obs/): the same captions as a transparent browser source for an OBS or vMix stream, with optional live translation (the original can be shown above it) and styling set from the app; no plugin. The page shows the overlay as the app draws it, with the controls to try.
- [Live translation for church services](https://speecho.app/live-translation-for-church/): the sermon as live translated subtitles on the screen or on visitors' phones through a QR code; no receivers, no app to install, billed by the second.
- Live interpreter by situation: [meetings in Google Meet, Zoom and Teams](https://speecho.app/live-interpreter-for-meetings/) (browser tab audio, one way, no bot), [live video translation](https://speecho.app/live-video-translation/) (webinars, courses, the original kept quietly underneath), [tour guides](https://speecho.app/live-interpreter-for-tour-guides/) (a printed QR code instead of receivers, up to 20 guests). Each also in es, de, fr, pt, ja, it, nl.
- Live translation pages for specific language pairs, each opening the app with both languages preset: [English to Spanish](https://speecho.app/live-translation/english-to-spanish/), [Spanish to English](https://speecho.app/live-translation/spanish-to-english/), [English to Polish](https://speecho.app/live-translation/english-to-polish/), [German to Vietnamese](https://speecho.app/live-translation/german-to-vietnamese/), [Portuguese to English](https://speecho.app/live-translation/portuguese-to-english/), [Tamil to English](https://speecho.app/live-translation/tamil-to-english/), [Polish to Ukrainian](https://speecho.app/live-translation/polish-to-ukrainian/), [German to Ukrainian](https://speecho.app/live-translation/german-to-ukrainian/), [English to Arabic](https://speecho.app/live-translation/english-to-arabic/) (right-to-left subtitles), [English to Vietnamese](https://speecho.app/live-translation/english-to-vietnamese/), [German to Arabic](https://speecho.app/live-translation/german-to-arabic/), [Korean to English](https://speecho.app/live-translation/korean-to-english/). Two of these also exist written in the audience's own language: [Deutsch-Ukrainisch](https://speecho.app/de/live-translation/deutsch-ukrainisch/) and [inglés-español](https://speecho.app/es/live-translation/ingles-espanol/). Any recognised language can be translated into any supported target, with or without a page.
- [PDF to audio](https://speecho.app/pdf-to-audio/): open a PDF, Word (DOCX) or EPUB file and get it read aloud in a natural voice as one MP3; the text is extracted in the browser for free, a long document is voiced in parts with the first part heard before the rest is paid for, up to 250,000 characters.
- [Ebook to audiobook](https://speecho.app/ebook-to-audiobook/): open an EPUB, Word (DOCX) or PDF book and get an M4B audiobook file with chapters, the title and the book's cover for Apple Books and audiobook apps, or one MP3 per chapter named after it; chapters come from the book's table of contents, heading styles or PDF bookmarks, can be left out before voicing, and cost nothing extra. Up to 250,000 characters (almost four hours of speech) per pass: a longer book is voiced in passes, and the M4B joins the passes into one file. Already have the audio? The free MP3 to M4B tool packs it. Also in es, de, fr, pt, ja, it, nl.
- [Text to podcast](https://speecho.app/text-to-podcast/): an article, PDF chapter or notes (300 to 60,000 characters) turned into a conversation between two AI hosts of about 5, 10 or 20 minutes, in the text's language or another; the hosts keep to what the text says, the script opens in the Dialogue editor to check and edit, then it is voiced into one MP3. The script costs 15, 25 or 40 credits for up to 5, 10 or 20 minutes plus 1 credit per 3,000 characters of text (a short text is not stretched and pays for the script it gets), the voicing the same as any dialogue. Also in es, de, fr, pt, ja, it, nl.
- [Bulk text to speech](https://speecho.app/bulk-text-to-speech/): a word list or a set of short phrases voiced line by line, every line its own MP3, in one voice and one set of settings; the same list can be voiced in up to three voices at once, each its own list and ZIP, priced per voice. Paste up to 1,000 lines (250,000 characters) or open a .txt, .csv or .tsv file (a file with a second column of translations becomes flashcards, below); the files are named after their lines and numbered in order ("0001_word.mp3") and download together as one ZIP. Chosen lines can be voiced again, or corrected first, each at its own characters. The list is priced as one text at the usual rates (50 / 110 / 160 credits per 1,000 characters), with one minimum for the whole list and every line counted as at least 10 characters: about 1,600 credits for 1,000 short words on Premium. In the Voice tab as List mode. Also in es, de, fr, pt, ja, it, nl.
- [Flashcard audio](https://speecho.app/flashcard-audio/): a word list with translations voiced as flashcards: each word in a native voice of its language, then after a pause of 0.5, 1 or 2 seconds its translation in a voice of the other language. Cards come from a two-column CSV/TSV (a header row is left out), two columns pasted from a spreadsheet, or lines written as word → translation; up to 1,000 cards (250,000 characters on both sides). One MP3 per card, or two files per card (the word, and the translation as _2), named after the word and numbered, in one ZIP, plus a CSV for Anki whose [sound:...] fields name the files. Priced as a list of all the sides (each side counted as at least 10 characters): about 3,200 credits for 1,000 short cards on Premium. In the Voice tab as List mode. Also in es, de, fr, pt, ja, it, nl.
- [Video translator](https://speecho.app/video-translator/): a video translated into another language with a voice-over that keeps the original timing. The video is uploaded in Transcribe (the browser takes the sound out and uploads only that), transcribed (99 languages), translated (18 languages; the first two translations of a transcript are free), then "Voice on the original timing" in the Voice menu voices every line at the second it was said, as one MP3 as long as the video. Several people talking get a voice each (up to 8; the panel asks to check who says each line, as that comes from speech recognition). A line longer than its time is sped up by up to 25% with the pitch kept, past that the next lines start a little late; the card shows which. Any line can be corrected or given another voice and voiced again alone. Not lip sync; the voice track holds only the new voice. Up to 2 hours of video for now. Price: transcription 10 credits a minute, voice 50 / 110 / 160 credits per 1,000 characters (about 900 characters a minute of steady narration), so about 1,100 credits for a 10-minute video on Enhanced. The track goes on the video in the free replace-audio tool below. Also in es, de, fr, pt, ja, it, nl.
- [Subtitles to voice-over](https://speecho.app/subtitles-to-voiceover/): an SRT, WebVTT or SBV file (opened or pasted) voiced in Text to speech's Subtitles mode, every subtitle at its own time, as one MP3 track, each subtitle also as its own file in a ZIP named by its time. Tags such as <i> are dropped, a subtitle of only a sound tag ([music], ♪) stays silent and is not charged, WebVTT voice tags (<v Name>) give each named speaker a voice (up to 8). Subtitles likely too long for their time are named before voicing. Up to 1,000 subtitles and 2 hours. Priced as one text at the usual rates, each subtitle counted as at least 10 characters. Also in es, de, fr, pt, ja, it, nl.
- [Handwriting to text](https://speecho.app/handwriting-to-text/): photograph handwritten pages (notes, letters, journals, cursive) and get editable text; AI handwriting OCR, 40 credits per page.
- More feature pages: [Chat with your audio](https://speecho.app/chat-with-audio/) (ask questions about a recording in plain language; the answers come from its transcript), [AI chapter generator](https://speecho.app/ai-chapter-generator/) (titled chapters with timestamps in YouTube format, free with every transcription), [AI voiceover for videos](https://speecho.app/ai-voiceover-for-videos/) (a script to a narration MP3), [Text to MP3](https://speecho.app/text-to-mp3/), [Phone greeting generator](https://speecho.app/phone-greeting-generator/) (voicemail and IVR prompts as MP3).

## By recording source and file format
One page per source or format, each saying how to get the file and what comes back (a transcript with speaker labels and timestamps, subtitles, an AI summary). Each also in es, de, fr, pt, ja, it, nl.
- Meeting recordings: [Zoom](https://speecho.app/zoom-recording-to-text/), [Google Meet](https://speecho.app/google-meet-recording-to-text/), [Microsoft Teams](https://speecho.app/teams-recording-to-text/), [Webex](https://speecho.app/webex-recording-to-text/), [Loom](https://speecho.app/loom-to-text/), and [any of them](https://speecho.app/for-meeting-transcription/) without a bot in the call.
- Files: [MP3](https://speecho.app/mp3-to-text/), [M4A](https://speecho.app/m4a-to-text/), [WAV](https://speecho.app/wav-to-text/), [MP4](https://speecho.app/mp4-to-text/), [MOV](https://speecho.app/mov-to-text/), [WebM](https://speecho.app/webm-to-text/), [voice memos](https://speecho.app/voice-memo-to-text/) from an iPhone, an Android phone or WhatsApp.

## By audience
The same product described for one kind of work at a time. Each also in es, de, fr, pt, ja, it, nl.
- [Journalists](https://speecho.app/for-journalists/) (source interviews and press conferences), [researchers](https://speecho.app/for-researchers/) (qualitative interviews and focus groups, exported for coding in NVivo, MAXQDA or Dedoose), [students](https://speecho.app/for-students/) (lectures into notes), [lawyers](https://speecho.app/for-lawyers/) (depositions, witness interviews, client meetings), [recruiters](https://speecho.app/for-recruiters/) (screening calls, a summary per candidate), [sales teams](https://speecho.app/for-sales-teams/) (calls and demos, action items and objections), [coaches](https://speecho.app/for-coaches/) (session notes), [podcasters](https://speecho.app/for-podcasters/) (show notes, chapters, translations), [YouTubers](https://speecho.app/for-youtubers/) (captions, descriptions, chapter outlines).
- [Voice messages](https://speecho.app/for-voice-messages/) (WhatsApp, Telegram and team voice notes, a whole stack at once) and [recording in the browser](https://speecho.app/for-voice-recorder/) (record on a phone or laptop with nothing to install, text in seconds).