feat: supertonic TTS shared workflow + daily-news voice artifact

[Copilot]

Wraps the supertone-inc/supertonic-py on-device TTS SDK as a reusable shared workflow and wires it into daily-news to produce a spoken digest uploaded as a run artifact.

shared/supertonic.md (new)

• Pre-agent steps: installs supertonic[serve], starts HTTP server on 127.0.0.1:7788, polls readiness up to 3 min (first run downloads ~400 MB model from HuggingFace)
• Agent prompt: full curl reference for both endpoints, voice/language/quality tables, artifact upload walkthrough

# native endpoint
curl -X POST http://127.0.0.1:7788/v1/tts \
  -H 'content-type: application/json' \
  -d '{"text":"...", "voice":"F1", "lang":"en", "steps":8}' \
  -o output.wav

# OpenAI-compatible alias
curl -X POST http://127.0.0.1:7788/v1/audio/speech \
  -H 'content-type: application/json' \
  -d '{"model":"supertonic-3", "input":"...", "voice":"F1", "response_format":"wav"}' \
  -o output.wav

daily-news.md (updated)

• Imports shared/supertonic.md
• timeout-minutes 30 → 50 to cover model download + synthesis
• Adds 🔊 Voice Summary phase: agent composes a 2–4 sentence spoken digest, synthesizes WAV via the local server (jq-safe quoting to avoid shell escaping issues), uploads via upload_artifact safe-output, and embeds the download link in the discussion

[github-actions]

Hey @copilot-swe-agent 👋 — great work adding the supertonic TTS shared workflow and wiring it into daily-news! The PR description is thorough and the split between shared/supertonic.md and the daily-news.md integration is clean.

One thing that would help before merging:

• Add a smoke-test step — there are no test files in this diff. For a shared workflow that downloads a ~400 MB model and starts an HTTP server, a minimal validation step (e.g., a dedicated test workflow that calls shared/supertonic.md, synthesises a single short phrase, and checks the WAV file size is non-zero) would catch regressions and give reviewers confidence the pre-agent boot sequence actually works end to end.

If you would like a hand, you can assign this prompt to your coding agent:

Add a smoke-test workflow for the new supertonic shared workflow.

Create `.github/workflows/test-supertonic.md` that:
1. Imports `shared/supertonic.md` to start the TTS server.
2. Synthesises a short phrase (e.g. "Hello world") using the native `/v1/tts` endpoint and saves it to `test-output.wav`.
3. Asserts the output file exists and is larger than 1 KB (to confirm actual audio was produced, not an empty file).
4. Optionally uploads `test-output.wav` as a run artifact for manual spot-checking.

Also update `daily-news.lock.yml` if the lock file needs to reflect the new test workflow.

Generated by ✅ Contribution Check · 680.7 AIC · ⌖ 13.3 AIC · ⊞ 24.7K · ◷