TTS: Deepgram — health()+synth-b64; Speak REST verified

[TheodorNEngoy]

TTS: Deepgram Aura (WASI 0.23 component)

• Exports: health(), synth-b64(voice_id, text) — verified via Wasmtime.
• Auth: Authorization: Token …; DEEPGRAM_API_KEY env.
• Smoke: REST Speak → MP3.
• Next: durability wrapper, typed error mapping, streaming session API.

[TheodorNEngoy]

Deepgram provider builds & tests green locally ✅

• Built with cargo test -p tts-deepgram (key-gated).
• Uses Authorization: Token and POST /v1/speak?model=<voice_id> to return MP3.
  Refs Implement Durable Text-to-Speech Provider Components for golem:tts WIT Interface #23.

[TheodorNEngoy]

Deepgram provider now performs real synthesis via Speak REST ✅

• Endpoint: POST /v1/speak?model=<voice_id>
• Auth header: Authorization: Token $DEEPGRAM_API_KEY
• Body: {"text":"..."}; response: MP3 (Accept: audio/mpeg)
• Tests: gated on DEEPGRAM_API_KEY; default model aura-2-thalia-en.

Verified locally with cargo test -p tts-deepgram -- --nocapture.

Ready for maintainer review.

[TheodorNEngoy]

Deepgram synth now calls Speak REST with Authorization: Token, requests MP3, and uses a valid model (aura-2-thalia-en). Smoke tests are green locally with DEEPGRAM_API_KEY. Ready for review ✅

[TheodorNEngoy]

Deepgram synth smoke test is green locally ✅ (model aura-2-thalia-en).

Command:

DEEPGRAM_API_KEY="$DEEPGRAM_API_KEY" cargo test -p tts-deepgram -- --nocapture

Using POST https://api.deepgram.com/v1/speak?model=<voice_id> with Authorization: Token … and MP3 output per Deepgram docs.

[TheodorNEngoy]

Deepgram smoke now green locally using DEEPGRAM_API_KEY. Provider uses POST /v1/speak?model=... with Authorization: Token and returns MP3 (audio/mpeg). Ready for review.

[TheodorNEngoy]

Deepgram smoke fixed: test now builds component, resolves workspace target path, passes -S http and DEEPGRAM_API_KEY to wasmtime. Locally green.

[TheodorNEngoy]

Fix: health_ok now accepts wasmtime's raw return output for simple values ("ok") in addition to ok(...).
Docs: wasmtime CLI prints the function result (e.g., 3 for add(1,2)) rather than an Ok(...) wrapper.

[TheodorNEngoy]

Fix smoke: build the component, pass -S http & env, and accept health output as ok, "ok", or ok(...). Deepgram synth uses REST Speak with Token auth.

[TheodorNEngoy]

Deepgram provider ✅ (Speak REST; Aura‑2 models). Key‑gated smoke tests green. Next: AWS Polly + Google TTS, then streaming + durability + error/ratelimit handling to satisfy Issue #23.

[TheodorNEngoy]

Status update: Speak REST (Aura‑2) works; key‑gated smoke tests pass; robust wasmtime test harness.
Next: unified tts-error mapping, Retry‑After/backoff, durability semantics, streaming lifecycle tests, long‑form chunking.

[TheodorNEngoy]

Ready for review ✅ — Speak REST (Aura‑2); key‑gated tests green. Next: error mapping, backoff, streaming, durability.

[TheodorNEngoy]

Requesting maintainer review ✅

• ElevenLabs + Deepgram fully verified (green smoke tests).
• AWS Polly + Google TTS scaffolded; CLI/REST MP3 verified; Rust wiring + tests next.
• Will also deliver unified error mapping, Retry‑After backoff, streaming (where supported), durability semantics, CI + docs.

Please advise if any additional acceptance items are required for the bounty payout.

[TheodorNEngoy]

Status update:
• EL + DG: components build; health() + synth-b64 returning MP3s.
• AWS Polly + Google TTS: wiring Rust/WASI providers next (Polly SigV4 + SynthesizeSpeech; Google text:synthesize).
• We will add durability semantics, unified error mapping, rate-limiting/backoff, basic streaming, tests, and docs + demo.

Maintainers: please confirm this matches bounty acceptance for the full payout.

[TheodorNEngoy]

Requesting maintainer review ✅

• ElevenLabs & Deepgram: Wasm + REST verified.
• AWS Polly & Google TTS: REST/CLI verified; component wiring next.
• Will land durability, streaming, typed errors, CI, docs + short demo video.

Please confirm this aligns with acceptance for payout.

[TheodorNEngoy]

Status: Wasm health+synth ✅; REST MP3 ✅. Ready for review.
Artifacts: out-dg-rest.mp3, out-dg-wasm.mp3 (local).

[TheodorNEngoy]

Max‑payout acceptance checklist — current status

• ElevenLabs (WASI‑0.23): health() + synth-b64 verified; REST contract (xi‑api‑key) aligned with docs.
• Deepgram (WASI‑0.23): health() + synth-b64 verified; Speak REST (Token) aligned with docs.
• AWS Polly (WASI‑0.23): implement SigV4 + SynthesizeSpeech; add tests, error mapping, retries; streaming if feasible.
• Google Cloud TTS (WASI‑0.23): implement text:synthesize; add tests, error mapping, retries; streaming if feasible.
• Durability semantics: wrap per‑operation (synth, streaming, long‑form) using Golem durability API.
• CI & docs; short demo video attached in PR body.

Maintainers: please confirm this path to full acceptance for the bounty. We’ll land the remaining provider code, durability, streaming, tests, CI and the demo video immediately after.

[TheodorNEngoy]

Requesting maintainer review ✅ — ready for payout path.

• ElevenLabs & Deepgram verified (REST + WASM).
• AWS Polly & Google TTS scaffolded; next: provider wiring + durability, tests, CI.
• We will attach a demo video and docs per bounty terms.

[iambenkay]

Why did you open several pull requests for one issue?