How to Transcribe Audio with Bitcoin Lightning
Upload audio, pay a Lightning invoice, get your transcript. Word-level timestamps, speaker labels, 13 languages.
For Humans
Use the web UI — no setup required.
Transcription services charge $10–25/month or $0.25/minute and require an account before you can upload a single file. At 10 sats/minute (~$0.004), a 1-hour recording costs ~600 sats — about $0.24. Upload the file, pay the invoice, get your transcript. No account, no subscription, nothing stored.
Meeting and interview notes
Upload a recording and get a timestamped transcript with speaker labels. Pass it to AI Chat to extract action items and summaries automatically.
Subtitles and captions
Download your transcript as an .srt file for direct use in video editors. Accurate word-level timestamps make it easy to sync captions to any video.
Multilingual transcription
13 languages supported including Spanish, French, Chinese, Arabic, and Japanese, with auto-detection if you're not sure. All at the same per-minute rate.
Sensitive recordings
No account means no file history linked to your identity. Pay a Lightning invoice and upload — nothing tied to you, nothing stored after delivery.
Available Model
Voxtral Mini Transcribe V2
For Agents
Integrate via L402, MCP, or OpenClaw — no account, no API key.
The transcription endpoint supports L402 authentication: your agent uploads an audio file, pays a Lightning invoice priced by duration, and receives the transcript JSON. Word-level timestamps and speaker labels included. Flat cost per minute, no monthly quota.
Automated meeting intelligence
An agent transcribes a recorded meeting, passes the transcript to AI chat for summarization, then sends the summary by email — fully automated, no human review needed.
Subtitle generation pipelines
A video processing agent transcribes each upload and generates an .srt file automatically — no manual captioning, billed only for what you process.
Searchable audio archives
Batch-transcribe a library of recordings to make them full-text searchable. Speaker labels and timestamps mean you can jump directly to any moment.
Transcribe then translate
An agent transcribes in the source language, passes the text to AI chat for translation, then re-voices it with TTS — full localization pipeline.
L402 Authentication Flow
Send the request without auth
POST your audio file and options (language, timestamps, diarization) to the endpoint. The server responds with HTTP 402 + a Lightning invoice priced by file duration.
Pay the Lightning invoice
Your agent pays the invoice using a Lightning wallet or library. Save the preimage from the payment result.
Resend with the preimage
Repeat the identical request, adding an Authorization: L402 <macaroon>:<preimage> header. The server returns the transcript JSON with timestamps and speaker labels.
# One command — lnget handles invoice, payment, and retry automatically
lnget -X POST https://sats4ai.com/api/l402/transcribe-audio \
--json '{"file":"data:audio/mp3;base64,<BASE64_AUDIO>","type":"transcription","language":"en"}'
# → { "text": "..." }Connect any MCP-compatible AI assistant. Transcription is one of 10+ tools available through a single connection.
View MCP docs →One-line MCP setup. Your agent transcribes with Lightning — no API key, no billing account.
Setup OpenClaw →Chain with Other Services
Try It Now — No Signup Required
Upload your audio, pay with Lightning, get your transcript. 10 sats per minute.