# Audio & Video Transcription to Text, SRT, VTT (`bite-sized-insights/transcription`) Actor Turn audio and video links into text with timestamped segments. Get optional SRT or WebVTT subtitles, automatic or chosen language detection from 100+ languages, and optional English translation. Process up to 100 files per run. - **URL**: https://apify.com/bite-sized-insights/transcription.md - **Developed by:** [bite-sized-insights](https://apify.com/bite-sized-insights) (community) - **Categories:** AI, Videos, For creators - **Security:** Limited permissions - **Stats:** 3 total users, 2 monthly users, 100.0% runs succeeded, 0 bookmarks - **User rating**: No ratings yet ## Pricing from $0.01 / 1,000 row returneds This Actor is paid per event and usage. You are charged both the fixed price for specific events and for Apify platform usage. Learn more: https://docs.apify.com/actors/running/actors-in-store.md#pay-per-event ## What's an Apify Actor? An Actor is a serverless cloud program that runs on the Apify platform. It has two run modes. In Batch mode, an Actor accepts a well-defined JSON input, performs an action which can take anything from a few seconds to a few hours, and optionally produces a well-defined JSON output, datasets with results, or files in key-value store. In Standby mode, an Actor provides a web server which can be used as a website, API, or an MCP server. Apify vocabulary and the platform model are defined once, in the agent quickstart at https://apify.com/agents.md. ## How to integrate an Actor? If asked about integration, you help developers integrate Actors into their projects. You adapt to their stack and deliver integrations that are safe, well-documented, and production-ready. Do not guess an integration path. Every one of them is in the agent quickstart at https://apify.com/agents.md: the Apify MCP server, Agent Skills with the Apify CLI, the JavaScript and Python clients, the REST API, and the account-free path for an agent with no human to sign in. It also carries the rule on stating cost before the first paid run. For examples already wired to this Actor's own input schema, see the [API](#api) section below. Each client library has reference documentation the quickstart does not restate: [JavaScript/TypeScript](https://docs.apify.com/api/client/js/docs.md) (`npm install apify-client`) and [Python](https://docs.apify.com/api/client/python/docs.md) (`pip install apify-client`). # README ## Audio & Video Transcription - Speech to Text, SRT & VTT Turn audio and video links into text. Give it a list of direct links to recordings, podcasts, calls, lectures or videos; every file comes back as one dataset row with the full transcript, timestamped segments, optional subtitle files and an optional English translation. Billed per audio minute, only for the minutes that were transcribed. ### What this actor does - **Transcribes the files you actually have**: MP3, M4A, MP4, MOV, WAV, OGG/Opus, FLAC, AAC, WebM, MKV, AVI, FLV, AIFF, AMR, WMA and MPEG-TS. The type is detected from the file itself, so a wrong file extension or content type does not matter. For video files the audio track is used. - **Timestamps**: every row carries the transcript as segments with start and end time in seconds, ready for search, highlights, chapters or alignment with the video. - **Subtitles**: ask for SRT and/or WebVTT and get ready-to-use subtitle text in the same row. - **100 languages**: the spoken language is detected automatically, or you pass it as a hint if you know it. Accuracy is best for widely spoken languages and lower for rare ones. - **Translate to English**: switch on translation and get the English transcript of speech in any supported language, at the same price. - **Safe by design**: only public http(s) links are fetched, files are limited to 300 MB, and the minute limit keeps a single huge recording from running up your bill. Direct media links only. A link to a web page that merely plays a video (a streaming or social media page) is not a media file and is reported as unsupported. ### Quick start ```json { "urls": ["https://upload.wikimedia.org/wikipedia/commons/d/dd/Armstrong_Small_Step.ogg"], "subtitleFormats": ["srt"] } ``` Click **Start**. Each file is added to the dataset as soon as it is transcribed. ### Input parameters | Parameter | Type | Required | Description | |---|---|---|---| | `urls` | Array\[String] | Yes | Links to the audio or video files. Up to 100 per run, up to 2048 characters each. | | `language` | String | No | `auto` (default) detects the language; or a code such as `en`, `de`, `es`, `ja`. | | `translateToEnglish` | Boolean | No | Return the English translation of the speech. Off by default. | | `subtitleFormats` | Array\[String] | No | Any of `srt` and `vtt`. Empty by default (no subtitle text). | | `maxMinutesPerFile` | Integer | No | Transcribe only the first N minutes of each file. 1-120, default 60. | | `withSummary` | Boolean | No | Add a compact summary of each transcript. Off by default. | | `withTags` | Boolean | No | Add single-word topic tags for each transcript. Off by default. | | `with5w1h` | Boolean | No | Add who, what, when, where, why and how of each transcript. Off by default. | | `withSentiment` | Boolean | No | Add the overall sentiment of each transcript. Off by default. | Entries without a scheme are treated as `https://`. Links to private or local addresses, links with credentials and malformed links are rejected as invalid input before anything is fetched. ### Output One row per input file. Every key is always present: missing values are `null`, lists are empty, never omitted. #### Transcribed file ```json { "url": "https://example.com/podcast.mp3", "status": "success", "language": "en", "language_probability": 0.99, "translated": false, "duration_seconds": 84.3, "transcribed_seconds": 84.3, "minutes_transcribed": 2, "truncated": false, "text": "Welcome to the show. Today we talk about timestamps.", "segments": [ {"id": 0, "start": 0.0, "end": 2.4, "text": "Welcome to the show."}, {"id": 1, "start": 2.9, "end": 6.1, "text": "Today we talk about timestamps."} ], "srt": "1\n00:00:00,000 --> 00:00:02,400\nWelcome to the show.\n\n2\n00:00:02,900 --> 00:00:06,100\nToday...", "vtt": null, "error_code": null, "error": null, "summary": { "summary": "A short welcome and an introduction to timestamps.", "stats": {"original_length": 52, "summary_length": 50, "compression_ratio": 0.9615} }, "tags": {"tags": ["podcast", "timestamps"], "count": 2}, "five_w1h": null, "sentiment": {"sentiment": "positive", "confidence": 0.82}, "analysis_truncated": false } ``` #### Failed file ```json { "url": "https://example.com/page.html", "status": "failed", "language": null, "language_probability": null, "translated": false, "duration_seconds": null, "transcribed_seconds": 0.0, "minutes_transcribed": 0, "truncated": false, "text": null, "segments": [], "srt": null, "vtt": null, "error_code": "unsupported_format", "error": "Unsupported file type. Supported: MP3, M4A/MP4, WAV, OGG/Opus, FLAC, AAC, WebM/MKV and more.", "summary": null, "tags": null, "five_w1h": null, "sentiment": null, "analysis_truncated": false } ``` The add-on fields (`summary`, `tags`, `five_w1h`, `sentiment`) are `null` unless you switched the add-on on and it succeeded; `five_w1h` also stays `null` for transcripts shorter than 50 characters. `analysis_truncated` is `true` when a transcript was longer than 100,000 characters (roughly 120 minutes of speech at a normal pace): the add-ons then only analysed its beginning. `status` is one of: - `success` - the whole file (or the first `maxMinutesPerFile` minutes of it) was transcribed - `partial` - the time budget for the file ran out; `segments` holds everything transcribed until then. If the budget runs out before the first segment is finished, the row is `failed` with `transcription_failed` instead - `failed` - nothing was delivered; `error_code` and `error` say why (no transcription charge) `error_code` values: `blocked_url` (private or unresolvable address), `download_failed`, `too_large` (over 300 MB), `unsupported_format` (not a supported audio or video file), `invalid_media` (corrupt or unreadable), `no_audio` (a video without an audio track), `too_long` (over 12 hours), `no_speech` (no speech found), `transcription_failed`, and `service_unavailable` (the transcription service could not be reached; try again). ### Pricing: pay per audio minute | Event | Price | |---|---| | Audio minute transcribed | $0.010 (**$10.00 per 1,000 minutes**) | | Summary add-on, per transcribed minute | $0.002 | | 5W1H add-on, per transcribed minute | $0.002 | | Tags add-on, per transcribed minute | $0.001 | | Sentiment add-on, per transcribed minute | $0.0005 | A flat start fee of $0.00005 per run applies, and Apify's platform dataset-item fee ($0.00001 per row) applies to every row written to the dataset, failed rows included. Minutes are counted per file and rounded up: a 61-second file is 2 minutes. Only minutes that were actually transcribed are billed: failed files, files without speech and translation requests cost nothing extra, and a file cut short by `maxMinutesPerFile` or by the time budget is billed for the audio that was transcribed. `minutes_transcribed` in every row shows exactly what was billed. Add-ons are optional and only billed when switched on: each is charged with the same minute count as the transcription, for every file where that add-on succeeded. If an add-on fails the transcript is still delivered, the add-on field is `null` and it is not billed. #### Free test runs New users on Apify's free plan get up to 5 free runs of this actor (lifetime total), each capped to 3 files of 2 minutes. Runs started via the Apify Console's "Test" button have the same caps regardless of plan and do not count against the 5 free runs. ### How long does a run take? Files are transcribed two at a time and results appear in the dataset as each one finishes. Expect roughly 8 to 12 times real time, so a one-hour recording takes about 5 to 8 minutes. Each file has a time budget of about 15 minutes. On a busy or slow server even a 120-minute recording can run out of time and come back as `partial` with the minutes transcribed so far. Lower `maxMinutesPerFile` or split the recording to avoid that. ### Use it from code Replace `` with your API token from the Apify Console (Settings, API and integrations). ```bash curl -X POST "https://api.apify.com/v2/acts/7V0WYMgIcZ4QCCPYO/runs?token=" \ -H "Content-Type: application/json" \ -d '{"urls": ["https://example.com/podcast.mp3"], "subtitleFormats": ["srt"]}' ``` The call returns the run immediately; read the results from the run's default dataset when it has finished. ```python from apify_client import ApifyClient client = ApifyClient("") run = client.actor("7V0WYMgIcZ4QCCPYO").call(run_input={ "urls": ["https://example.com/podcast.mp3"], "language": "auto", "subtitleFormats": ["srt", "vtt"], }) for item in client.dataset(run["defaultDatasetId"]).iterate_items(): print(item["status"], item["minutes_transcribed"], item["text"]) ``` ```javascript import { ApifyClient } from 'apify-client'; const client = new ApifyClient({ token: '' }); const run = await client.actor('7V0WYMgIcZ4QCCPYO').call({ urls: ['https://example.com/podcast.mp3'], subtitleFormats: ['srt'], }); const { items } = await client.dataset(run.defaultDatasetId).listItems(); console.log(items[0].status, items[0].text); ``` Use the asynchronous run shown above for long recordings. Apify's synchronous run endpoints stop waiting after about five minutes, which is shorter than the time a long file can need. ### Use with AI agents (MCP) Connect an AI client such as Claude Desktop, Cursor or VS Code to the Apify MCP server and it can call this actor as a tool. Authenticate with OAuth or your API token: ```json { "mcpServers": { "apify-transcription": { "url": "https://mcp.apify.com/?actors=7V0WYMgIcZ4QCCPYO", "headers": {"Authorization": "Bearer "} } } } ``` Ask the agent to transcribe a direct media link; it passes the same input as in the table above and gets the dataset rows back. Billing is the same as for a normal run. ### Troubleshooting | Symptom | Cause | Fix | |---|---|---| | `blocked_url` | Private or local address, or the host name does not resolve | Use a public link | | `download_failed` | Server error, too many redirects, slower than 120 s, or expired link | Use a stable public URL | | `unsupported_format` | A web page, playlist or other non-media file | Link the media file itself | | `invalid_media` | The file is empty, damaged or unreadable | Check that it plays; re-export it | | `too_large` | Over 300 MB | Extract the audio as MP3 or AAC, or split it | | `no_audio` | The video has no audio track | Nothing to transcribe | | `too_long` | Over 12 hours | Split the recording | | `no_speech` | Only music, silence or noise | No transcription charge; check the file | | `transcription_failed` | Model error, or no segment finished in time | Retry; lower `maxMinutesPerFile` | | `service_unavailable` | Transcription service unreachable | Try again later | | `partial` status | Time budget ran out part way | Lower `maxMinutesPerFile`; split it | | Only 3 files of 2 minutes done | Free plan and Console test runs are capped | Use a paid plan | | Wrong language or odd words | Detection guessed wrong on a short or noisy clip | Set `language` | ### FAQ #### Why is a minute billed when the last part of my file is silent? Minutes are counted over the audio that was processed, silence included. Files in which no speech is found at all are reported as `no_speech` and carry no transcription charge. #### Can I transcribe only the start of a long recording? Yes. `maxMinutesPerFile` transcribes the first N minutes. The row shows the real length in `duration_seconds` and `truncated: true` when audio was left out. #### Which languages are supported? 100 languages. The Spoken language dropdown in the input form lists all of them. Automatic detection works for most recordings; a fixed language helps with short or noisy clips. Quality varies: it is best for widely spoken languages and lower for rare ones. #### How do I get the translation instead of the original wording? Switch on `translateToEnglish`. `language` is then the language that was spoken and `translated` is `true`. Speech that is already English is returned as is. #### Can I paste a YouTube or other streaming page link? No. Only direct links to media files are fetched; pages that need scraping or sign-in are not supported. #### Does this actor use a proxy? No. It calls a dedicated transcription service, and the service fetches your public media links from its own server address, not from Apify's IPs or yours. Links that only work from an allowlisted IP, behind a login, or with expiring tokens bound to another address will fail with `download_failed`. Use a public link or a signed URL that is valid for anyone. ### Disclaimer Only transcribe recordings you are allowed to process. Accuracy depends on the audio: background noise, overlapping speakers, strong accents and music can reduce quality, and names and rare terms may be misspelled. Translations are machine translations. Speakers are not identified. *** **Need help?** Open an issue on the actor's repository or contact support through Apify Console. # Actor input Schema ## `urls` (type: `array`): MP3, M4A, MP4, WAV, OGG, FLAC, WebM and more. Public http(s) links, up to 300 MB each. ## `language` (type: `string`): Leave on automatic unless you know it. A fixed language helps with short or noisy clips. ## `translateToEnglish` (type: `boolean`): Return the transcript translated into English. The price per minute stays the same. ## `subtitleFormats` (type: `array`): Also return the transcript as subtitle text in these formats. ## `maxMinutesPerFile` (type: `integer`): Longer recordings are cut after this many minutes. Only transcribed minutes are billed. ## `withSummary` (type: `boolean`): Add a compact summary of each transcript. Billed per transcribed minute. ## `withTags` (type: `boolean`): Add single-word topic tags for each transcript. ## `with5w1h` (type: `boolean`): Extract the 5W1H facts of each transcript, each with a confidence. ## `withSentiment` (type: `boolean`): Add the overall sentiment (positive, neutral or negative) of each transcript. ## Actor input object example ```json { "urls": [ "https://upload.wikimedia.org/wikipedia/commons/d/dd/Armstrong_Small_Step.ogg" ], "language": "auto", "translateToEnglish": false, "subtitleFormats": [], "maxMinutesPerFile": 60, "withSummary": false, "withTags": false, "with5w1h": false, "withSentiment": false } ``` # Actor output Schema ## `results` (type: `string`): One row per input file: text, timestamped segments, subtitles and the minutes billed # API You can run this Actor programmatically using our API. Below are code examples in JavaScript, Python, and CLI, as well as the OpenAPI specification and MCP server setup. ## JavaScript example ```javascript import { ApifyClient } from 'apify-client'; // Initialize the ApifyClient with your Apify API token // Replace the '' with your token const client = new ApifyClient({ token: '', }); // Prepare Actor input const input = { "urls": [ "https://upload.wikimedia.org/wikipedia/commons/d/dd/Armstrong_Small_Step.ogg" ] }; // Run the Actor and wait for it to finish const run = await client.actor("bite-sized-insights/transcription").call(input); // Fetch and print Actor results from the run's dataset (if any) console.log('Results from dataset'); console.log(`💾 Check your data here: https://console.apify.com/storage/datasets/${run.defaultDatasetId}`); const { items } = await client.dataset(run.defaultDatasetId).listItems(); items.forEach((item) => { console.dir(item); }); // 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/js/docs ``` ## Python example ```python from apify_client import ApifyClient # Initialize the ApifyClient with your Apify API token # Replace '' with your token. client = ApifyClient("") # Prepare the Actor input run_input = { "urls": ["https://upload.wikimedia.org/wikipedia/commons/d/dd/Armstrong_Small_Step.ogg"] } # Run the Actor and wait for it to finish run = client.actor("bite-sized-insights/transcription").call(run_input=run_input) # Fetch and print Actor results from the run's dataset (if there are any) print(f"💾 Check your data here: https://console.apify.com/storage/datasets/{run.default_dataset_id}") for item in client.dataset(run.default_dataset_id).iterate_items(): print(item) # 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/python/docs/quick-start ``` ## CLI example ```bash echo '{ "urls": [ "https://upload.wikimedia.org/wikipedia/commons/d/dd/Armstrong_Small_Step.ogg" ] }' | apify call bite-sized-insights/transcription --silent --output-dataset ``` ## MCP server setup ```json { "mcpServers": { "apify": { "type": "http", "url": "https://mcp.apify.com/?tools=fetch-actor-details,bite-sized-insights/transcription" } } } ``` The hosted server signs you in with OAuth on first connect, so no API token belongs in this config. Clients without OAuth support can send an `Authorization: Bearer ` header instead, using a token from API & Integrations in Apify Console (https://console.apify.com/settings/integrations). ## OpenAPI specification Download the OpenAPI definition: https://api.apify.com/v2/actors/7V0WYMgIcZ4QCCPYO/builds/juHRY5Hzd05DUusur/openapi.json