Reference · Common questions

Common questions

Privacy, languages, plans, export formats, speaker labels, AI summaries, and the API — answered once.

01Privacy & data

Privacy & data

What happens to your audio, what we store, what we don't train on.

  1. 01Is my file deleted after transcription?
    Yes. Source audio and video files are permanently deleted from our servers within 24 hours of transcription completing. Transcripts and summaries stay in your account until you delete them.
  2. 02Do you train AI models on my audio?
    No. We do not use customer audio, video, or transcripts to train models — ours or any third party's. The ASR and LLM providers we route through (AssemblyAI for speech, OpenRouter providers for summary) are contractually bound to the same.
  3. 03Where is my data stored?
    Audio files and database are hosted on Hetzner data centres in Germany (Falkenstein and Nuremberg). Backups go to Cloudflare R2 (EU region). Data does not leave the EU for the storage layer. The ASR provider (AssemblyAI) processes audio in the United States.
  4. 04Is the service GDPR-compliant?
    Yes. We are a Brazilian-registered company with EU data hosting, an explicit DPA available on request, the right to data export and deletion at any time, and 24-hour automatic deletion of source audio. Email privacy@transcription.solutions for the DPA or a data export.
02Languages

Languages

What we transcribe, how detection works, what's accurate.

  1. 01How many languages do you support?
    100+ languages. The full list is at /languages. Tier-1 languages (English, Spanish, German, French, Portuguese, Italian, Dutch, Polish) deliver 95%+ accuracy on clear speech. Tier-2 (Russian, Japanese, Mandarin, Korean, Arabic, and 13 others) deliver production-grade accuracy.
  2. 02Do I need to set the language manually?
    No. Language is auto-detected from the first 30 seconds of audio. For very short clips where detection might be uncertain, you can force a specific language in the advanced settings.
  3. 03What about code-switching or multilingual recordings?
    The detector adapts segment by segment for recordings that switch language mid-session. For high-stakes multilingual content (research interviews, multilingual panels), pass the dominant language as a hint to improve accuracy on the secondary segments.
  4. 04Does accuracy vary by language?
    Yes. English, Spanish, German, French, Portuguese, Italian, Dutch, and Polish are at the top of the accuracy curve — typically 95%+ on clear speech. Tier-2 languages run a few percentage points lower. Endangered or low-resource languages may not be supported — check the /languages list before relying on a specific one.
03Plans & limits

Plans & limits

What's free, what's paid, what each tier covers.

  1. 01What's the difference between Free, Pro, and Business?
    Free — 30 minutes per month, up to 30 minutes per file, 100 MB per file, 5 URL imports per month. All export formats. No speaker labels, no AI summary. Pro ($19/mo) — 600 minutes per month, up to 10 hours per file, 2 GB per file, unlimited URL imports, speaker labels, AI summary, 20 concurrent jobs. Business ($49/mo) — 2,500 minutes per month, up to 10 hours per file, 5 GB per file, everything in Pro plus priority queue and 50 concurrent jobs.
  2. 02Can I upgrade or downgrade anytime?
    Yes. Plan changes take effect immediately for upgrades (you get the new limits on the next file) and at the next billing cycle for downgrades (so you don't lose access to features you've already paid for that month).
  3. 03What happens when I hit my monthly limit?
    On Free, additional jobs are rejected until the monthly reset. On Pro and Business, you can either upgrade, wait until the cycle reset, or pay overage at $0.04/min on Pro and $0.02/min on Business. Overage is opt-in — we don't auto-charge unless you enable it in billing.
  4. 04Is there an annual discount?
    Yes — 50% off when you pay annually. The Free tier stays free either way. Cancel anytime; if you cancel mid-year, you keep access until the period ends.
04Export formats & files

Export formats & files

What you can download, what formats we accept.

  1. 01What formats can I export to?
    All plans (including Free): TXT, SRT, VTT, DOCX. Pro and Business additionally get Markdown, PDF (rich layout), and JSON via the API. PDF requires WeasyPrint server-side; on rare cases when it's unavailable you'll see a graceful 503.
  2. 02What audio and video formats do you accept?
    Audio: MP3, WAV, M4A, OGG, FLAC, OPUS, AMR. Video: MP4, MOV, MKV, AVI, WebM. We extract the audio track from video files server-side — no conversion needed on your end. We also accept URLs from YouTube, Vimeo, TikTok, Instagram, Twitter/X, WhatsApp web exports, and 1,500+ other sources via yt-dlp.
  3. 03What's the maximum file size?
    Free: 100 MB and 30 minutes per file. Pro: 2 GB and 10 hours per file. Business: 5 GB and 10 hours per file. Whichever limit comes first applies. For larger archives, submit multiple files through the API.
  4. 04Can I edit the transcript before exporting?
    Yes. The transcript viewer supports inline edits, speaker renames, and re-segmentation. Changes save automatically and reflect in every export format. Speaker renames propagate across every turn.
05Speaker labels & AI features

Speaker labels & AI features

Diarization, summaries, and what each plan unlocks.

  1. 01Which plans include speaker labels?
    Pro and Business. Speaker diarization separates two or more voices in the same recording and labels each turn as Speaker 1, Speaker 2, etc. — click any label to rename to the actual person. Free does not include speaker labels.
  2. 02How accurate is speaker diarization?
    Two-speaker recordings (interviews, sales calls, one-on-one meetings) get the cleanest separation — typically 95%+ correct attribution on clear audio. Three to four speakers in a roundtable get 90%+. Six or more speakers, or recordings with significant overlap, may need a manual review pass on borderline turns.
  3. 03What does the AI summary include?
    Four fields: a 2-4 sentence summary of the conversation, key points (3-7 main takeaways), action items (anything discussed for follow-up — empty if none), and topic tags (3-8 short keywords for browsing and search). Available on Pro and Business.
  4. 04Can I rename speakers in the transcript?
    Yes. Click any speaker label in the transcript viewer and type the actual name. The rename propagates across every turn for that speaker in the same job. Available on Pro and Business.
06API & integrations

API & integrations

Programmatic access, webhooks, rate limits.

  1. 01Do you have a REST API?
    Yes. Full docs at /docs/api. Submit files for transcription, poll job status, fetch transcripts in any export format, and receive webhook callbacks when jobs complete. Authentication is via API key (per-key rate limits) or session JWT (browser only).
  2. 02How does the webhook work?
    Pass a `webhook_url` in your POST to /api/v1/jobs. When the job reaches a terminal state (done or failed), we POST to that URL with the job ID and status. Re-fetch the transcript via GET /api/v1/jobs/{id}/export?format=… in your chosen format. We retry webhook delivery up to 5 times with exponential backoff on non-2xx responses.
  3. 03Can I process files in bulk?
    Yes — submit each file as a separate job to /api/v1/jobs. We don't currently have a bulk endpoint that accepts a list, but the concurrency model handles it transparently: submit hundreds in parallel, and any over your plan's active-job cap are parked as `pending` and auto-promoted as slots free. Pro runs up to 20 concurrent, Business up to 50.
  4. 04What's the rate limit on the API?
    Per-API-key rate limits prevent runaway scripts: 60 submissions per minute on Free, 300 per minute on Pro, 1000 per minute on Business. These are submission limits — the actual processing throughput is governed by your plan's max_concurrent_jobs.

Still have a question?

Email support@transcription.solutions and a real person will reply within one business day. For privacy and data requests, use privacy@transcription.solutions.

Back to home →API documentation →Supported languages →