Audio prompt mode (listen and identify) #37

New issue

Open

opened 2026-09-26 13:58:40 +00:00 by belvedere · 0 comments

belvedere commented

2026-09-26 13:58:40 +00:00

Collaborator

Objective: Add listening practice, deferred until the core loop works.

Files:

Create: lib/first_thousand_words/audio/provider.ex, .../audio/tts_provider.ex
Modify: priv/static/ or object storage for cached clips

Steps:

Generate audio per word once, cache by (word_id, voice); never synthesise per request.
Store audio_url on Word.
Add direction: :listen cards.
Make the provider configurable; keep the abstraction thin.

Verify: a listen card plays cached audio for 10 words without regenerating.

Notes: explicitly deferred. Marked future because it adds a storage dependency and a
vendor decision that should not block v1.

**Objective:** Add listening practice, deferred until the core loop works. **Files:** - Create: `lib/first_thousand_words/audio/provider.ex`, `.../audio/tts_provider.ex` - Modify: `priv/static/` or object storage for cached clips **Steps:** 1. Generate audio per word once, cache by `(word_id, voice)`; never synthesise per request. 2. Store `audio_url` on `Word`. 3. Add `direction: :listen` cards. 4. Make the provider configurable; keep the abstraction thin. **Verify:** a listen card plays cached audio for 10 words without regenerating. **Notes:** explicitly deferred. Marked `future` because it adds a storage dependency and a vendor decision that should not block v1.

belvedere added this to the M4 — Practice experience milestone

2026-09-26 13:58:40 +00:00

belvedere added the

labels

2026-09-26 13:58:40 +00:00