weft transcript to podcast

The problem

Developers often have long transcripts from meetings or videos that no one has time to read. Turning these into digestible audio content usually requires manual scripting, expensive voiceover work, and complex API orchestrations.

What it does

• Analyzes full transcripts to extract 3-5 core ideas and material caveats.
• Writes a focused, 5-minute narrative script (650-750 words) with source attribution.
• Generates a playable MP3 file using high-quality voices like ElevenLabs via the Weft proxy.
• Produces comprehensive show notes with source links and timestamped claim references.
• Manages paid API calls through Weft, including cost estimation and idempotency handling.

Frameworks & tools

Weft SDK/MCP, BlockRun TTS API, ElevenLabs models, and standard HTTP clients for media handling.

Why this beats prompting it yourself

Prompting alone won't handle the multi-step workflow of text analysis, character-limited speech generation, and secure paid API proxying. This skill manages the specific math for cost-capping, handles Weft authentication headers, and ensures the script is optimized for natural speech without narrator stage directions.

Use cases

• Turn technical YouTube tutorials into quick commute-friendly audio summaries.
• Convert long meeting transcripts into executive audio briefings for the team.
• Create "listenable" versions of long-form articles or documentation.
• Automate internal daily standup audio recaps from text-based logs.

Known limitations

Requires a Weft connection with paid POST support. Does not support voice cloning, podcast feed publishing, or local TTS fallbacks. Maximum script length is 40,000 characters.