Build an AI Transcription Pipeline with Daytona

[M4dotsuki]

Summary

/claim #13

Adds a long-form Daytona guide for building a repeatable AI transcription pipeline with Sapat.

Included

• New guide covering Daytona workspace setup, Sapat installation, ffmpeg, provider configuration, required --api, quality settings, prompt hints, transcript correction, batching, and review checks
• New speech-to-text transcription definition
• Original SVG workflow diagram
• First-time contributor author profile for M4dotsuki
• Follow-up coverage for extending Sapat with another provider

Related upstream work

• Opened nibzard/sapat#10 to add a Deepgram transcription provider to Sapat.
• Updated this guide to document the Deepgram provider extension and its .env configuration.

Validation

• npx markdownlint-cli authors/m4dotsuki.md definitions/20260511_definition_speech_to_text_transcription.md guides/20260511_build_an_ai_transcription_pipeline_with_daytona_and_sapat.md
• git diff --check
• Installed Sapat locally from nkkko/sapat and verified sapat --help exposes the documented options, including required --api, --quality, and --correct
• For the Sapat provider PR: python -m compileall src, python -m pip install -e ., python -m sapat.script --help, Deepgram provider import verification, python -m unittest tests.test_deepgram, and python -m compileall src tests

Note: repository-wide markdownlint currently reports pre-existing issues in older files, so validation was scoped to the files changed by this PR.

[M4dotsuki]

Follow-up to strengthen the original bounty requirement:

I opened a Sapat upstream PR adding another supported transcription provider:

• nibzard/sapat#10 - Add Deepgram transcription provider

That PR adds --api deepgram, documents the required environment variables, and verifies the CLI exposes the new provider. This complements the guide here by addressing the issue request to expand Sapat beyond the existing OpenAI/Groq/Azure provider set.