docs(integrations): cleanup — legacy endpoints, stale latencies, fake voice IDs - #256
Open
abhishekmishragithub wants to merge 2 commits into
Open
docs(integrations): cleanup — legacy endpoints, stale latencies, fake voice IDs#256abhishekmishragithub wants to merge 2 commits into
abhishekmishragithub wants to merge 2 commits into
Conversation
… voice IDs
Surface 3 of the cleanup sprint. Drift across six integration pages
(dograh, pipecat, openwhispr, livekit, vercel-ai-sdk, openclaw).
Three classes of drift cleaned up:
1. **Legacy `/v1/pulse/get_text` endpoint** still referenced in five
integration pages. The canonical unified path is
`/waves/v1/stt/?model=pulse` (HTTP) and
`wss://.../waves/v1/stt/live?model=pulse` (WS).
- `integrations/dograh.mdx` — Notes section.
- `integrations/pipecat.mdx` — STT service description.
- `integrations/openwhispr.mdx` — How-it-works POST example.
- `integrations/livekit.mdx` — STT endpoint paragraph (line not
touched by in-flight PR #251, which fixes lines 76/86).
- `integrations/vercel-ai-sdk.mdx` — Streaming STT section
(line not touched by in-flight PR #251).
2. **Stale "64 ms" / "Sub-100 ms" latency claims** in five
integrations. Real number is ~150 ms streaming TTFT at 1
concurrency (~300 ms at 100) per Gaurav's PR #238 review.
For TTS it's ~200 ms TTFB at 40 concurrent per the Lightning v3.1
model-card benchmark.
- `integrations/pipecat.mdx` — STT description.
- `integrations/livekit.mdx` — `smallestai.STT` summary bullet.
- `integrations/vercel-ai-sdk.mdx` — Streaming STT paragraph.
- `integrations/openclaw.mdx` — Features list (both STT 64ms
and TTS sub-100ms bullets — replaced with canonical
benchmarks).
3. **Fake voice ID `sophia`** used as a default in two pages.
`sophia` is not in the Lightning v3.1 standard or Pro catalogs
(Pro has `sophie`, not `sophia`). Customers pasting these
examples would get an API error or wrong-voice output. Replaced
with `magnus` (real Lightning v3.1 standard voice).
- `integrations/vercel-ai-sdk.mdx` — 3 occurrences (`voice: 'sophia'`,
`voice: voice || 'sophia'`, etc.) → `magnus`.
- `integrations/livekit.mdx` — TTS `voice_id` row in param table:
`"sophia"` → `"magnus"` as the standard-pool default; also
expanded the language description to clarify the Pro pool's
`en` / `hi` / omit→`en+hi` behaviour (consistent with PR #251).
Avoided overlap with PRs #251 (livekit/vercel-ai-sdk lang enum) and
#252 (n8n/openclaw emotion-gender) by targeting different lines.
Verified via `fern check` (0 errors). 6 files changed, 13/13.
This was referenced Jun 19, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Surface 3 of the cleanup sprint. Three classes of drift across six integration pages — dograh, pipecat, openwhispr, livekit, vercel-ai-sdk, openclaw.
1. Legacy `/v1/pulse/get_text` endpoint in 5 integrations
The canonical unified paths are `/waves/v1/stt/?model=pulse` (HTTP) and `wss://.../waves/v1/stt/live?model=pulse` (WS). All five updated.
2. Stale "64 ms" / "Sub-100 ms" latency claims
Real STT number is ~150 ms streaming TTFT at 1 concurrency, ~300 ms at 100 (Gaurav's PR #238 review). Real TTS number is ~200 ms TTFB at 40 concurrent (Lightning v3.1 model-card benchmark). Aligned across pipecat, livekit, vercel-ai-sdk, openclaw.
3. Fake voice ID `sophia`
Not in either Lightning v3.1 standard or Pro catalogs (Pro has `sophie`, not `sophia`). Customers pasting these examples would get an API error or wrong-voice output. Replaced with `magnus` (real standard-pool voice).
Avoiding in-flight PR conflicts
Verification
Test plan
Sprint context
Surface 1 (STT) → #254. Surface 2 (TTS) → #255. Surface 4 (rendered API ref MDX) is next.
🤖 Generated with Claude Code