realtime-avatar 0.37.0 → 0.39.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +29 -6
- package/dist/{chunk-FFMZERDZ.js → chunk-PZNCGKOB.js} +1 -1
- package/dist/{chunk-RBK6EHVH.js → chunk-YBKE7SDI.js} +470 -144
- package/dist/express.d.ts +2 -2
- package/dist/express.js +2 -2
- package/dist/hono.d.ts +2 -2
- package/dist/hono.js +2 -2
- package/dist/index.d.ts +40 -5
- package/dist/index.js +1 -1
- package/dist/nextjs.d.ts +2 -2
- package/dist/nextjs.js +2 -2
- package/dist/server.d.ts +2 -2
- package/dist/server.js +1 -1
- package/dist/tanstack-start.d.ts +2 -2
- package/dist/tanstack-start.js +2 -2
- package/dist/{types-DVL5VX3N.d.ts → types-B8MMMQ_V.d.ts} +445 -118
- package/dist/{types-RQo4LwZ3.d.ts → types-CpfJZaxo.d.ts} +1 -1
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -259,6 +259,24 @@ and fetch, including a pending `next()`. The smaller of `timeoutMs` and `totalTi
|
|
|
259
259
|
covers the entire request/body, including pauses between iterations. **Never automatically
|
|
260
260
|
retry** an ambiguous paid request. Cancellation does not prove that billing stopped.
|
|
261
261
|
|
|
262
|
+
### Live speech session
|
|
263
|
+
|
|
264
|
+
`openSpeechSession(voiceId, { maxTextBytes, latency?, chunkLength?, language? }, { signal? })`
|
|
265
|
+
opens `GET /voices/{voiceId}/speech/live` as a WebSocket and resolves once the platform admitted
|
|
266
|
+
the session, so every refusal rejects the promise. Send `text(utterance, text)` (utterance 0 first,
|
|
267
|
+
at most `chunkLength` characters each), `flush(utterance)` to speak it and open the next, `end()`
|
|
268
|
+
and `cancel()`. Iterate it once for `started`, per-utterance `audio` (`sampleOffset` from the
|
|
269
|
+
utterance's start), `alignment` (replaces that utterance's snapshot, may arrive late) and
|
|
270
|
+
`utterance-end`, then one `completed` or `failed`. Frames are validated by the generated
|
|
271
|
+
`speechLiveClientMessageSchema` and `speechLiveServerMessageSchema` in both directions;
|
|
272
|
+
non-contiguous audio or a skipped utterance fails the session.
|
|
273
|
+
|
|
274
|
+
Billing is for the UTF-8 bytes sent: cancel, disconnect, `break` and abort still bill them, and a
|
|
275
|
+
provider failure refunds. Never retried. In Workers the upgrade is a `fetch` with the
|
|
276
|
+
`Authorization` header and a refusal is a `RealtimeAvatarHttpError`; in Node 22 it is the global
|
|
277
|
+
`WebSocket`, which reports no status, so a refusal is a `RealtimeAvatarError`. `timeoutMs` bounds
|
|
278
|
+
opening and `totalTimeoutMs`, if set, the whole session.
|
|
279
|
+
|
|
262
280
|
## Custom voice (webhook)
|
|
263
281
|
|
|
264
282
|
Speak a call in your own voice. Each sentence she is about to say is POSTed, signed, to an
|
|
@@ -482,6 +500,8 @@ rta.getConnectionHistory(sessionId) // bounded LiveKit observations; requires
|
|
|
482
500
|
rta.createSpeech(voiceId, { text, format?, speed?, language? }, { signal? })
|
|
483
501
|
// unreleased streaming candidate; generated SSE events, never retried
|
|
484
502
|
rta.createSpeechStream(voiceId, { text, speed?, language? }, { signal? })
|
|
503
|
+
// live session: text in while it is written, PCM + alignment per utterance; WebSocket, never retried
|
|
504
|
+
rta.openSpeechSession(voiceId, { maxTextBytes, latency?, chunkLength?, language? }, { signal? })
|
|
485
505
|
|
|
486
506
|
// pre-rendered video (MP4, no live call); server only, requires renders:write / renders:read
|
|
487
507
|
rta.createRender({ avatarId, voiceId? | voice?, lipGain?, shots }) // 202 queued; one Idempotency-Key across retries
|
|
@@ -492,7 +512,7 @@ rta.iterateRenders({ limit? }) // every render, paging handled
|
|
|
492
512
|
rta.getRenderAccess(renderId) // { id, url, expiresAt }; signed for an hour, renew when needed
|
|
493
513
|
|
|
494
514
|
// avatars
|
|
495
|
-
rta.
|
|
515
|
+
rta.createAnimaAvatar({ displayName, voice, settings?, metadata? }) // app-owned rig and rendering, explicit voice, no assets
|
|
496
516
|
rta.createAvatarFromImage({ displayName, imageUrl, motionPrompt?, voice? }) // platform-rendered video
|
|
497
517
|
rta.createAvatarFromVideo({ displayName, videoUrl, voice? }) // DEPRECATED — closed, 422
|
|
498
518
|
rta.listAvatars()
|
|
@@ -544,10 +564,10 @@ const voice: VoiceSpecInput = {
|
|
|
544
564
|
};
|
|
545
565
|
```
|
|
546
566
|
|
|
547
|
-
### Register an app
|
|
567
|
+
### Register an Anima avatar (your app owns the rig and the rendering)
|
|
548
568
|
|
|
549
569
|
```ts
|
|
550
|
-
const avatar = await rta.
|
|
570
|
+
const avatar = await rta.createAnimaAvatar({
|
|
551
571
|
displayName: "Rin",
|
|
552
572
|
voice: { provider: "fish", voice_id: "your-provider-voice-id", model: "s2-pro" },
|
|
553
573
|
metadata: { characterId: "your-character-id" },
|
|
@@ -556,18 +576,21 @@ const avatar = await rta.createLive2dAvatar({
|
|
|
556
576
|
const call = await rta.startCall({ avatarId: avatar.id, mode: "voice" });
|
|
557
577
|
```
|
|
558
578
|
|
|
559
|
-
Requires the platform's
|
|
579
|
+
Requires the platform's `anima` renderer. One `POST /avatars` sends `renderer: "anima"`,
|
|
560
580
|
`sourceKind: "image"` and an explicit voice selector; no assets, automatic casting, clips or
|
|
561
|
-
resting-loop generation are requested. Your app owns the rig. Source fields, motion prompts,
|
|
581
|
+
resting-loop generation are requested. Your app owns the rig and the rendering. Source fields, motion prompts,
|
|
562
582
|
renderer overrides and automatic voice descriptions are refused before network I/O.
|
|
563
583
|
`settings` and `metadata` are optional records and are omitted unless provided.
|
|
564
584
|
|
|
565
|
-
The exported `
|
|
585
|
+
The exported `CreateAnimaAvatarInput` and complete `RegisteredAvatar` response derive from
|
|
566
586
|
canonical generated schemas. The response is parsed once; malformed successful responses throw
|
|
567
587
|
`RealtimeAvatarError` without the rejected body. Existing image/video creation requests are
|
|
568
588
|
unchanged. Existing avatar methods retain `renderer`, defaulting older responses that omit it to
|
|
569
589
|
`"video"`.
|
|
570
590
|
|
|
591
|
+
`renderer: "anima"` replaced `live2d` in 0.38.0 with no alias: the platform refuses `live2d` and
|
|
592
|
+
SDK 0.37.0 and earlier cannot parse an `anima` avatar, so upgrade once the platform serves `anima`.
|
|
593
|
+
|
|
571
594
|
### Validate persisted grants and proxy requests
|
|
572
595
|
|
|
573
596
|
The server root (`realtime-avatar` or `realtime-avatar/server`) exports the existing
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { RealtimeAvatar, ROUTE_TIMEOUT_MS, proxyEndRequestSchema, proxyConnectRequestSchema, isQueued, RealtimeAvatarHttpError, RealtimeAvatarError } from './chunk-
|
|
1
|
+
import { RealtimeAvatar, ROUTE_TIMEOUT_MS, proxyEndRequestSchema, proxyConnectRequestSchema, isQueued, RealtimeAvatarHttpError, RealtimeAvatarError } from './chunk-YBKE7SDI.js';
|
|
2
2
|
|
|
3
3
|
// ../http-client/src/generated/error-semantics.ts
|
|
4
4
|
var ERROR_SEMANTICS = {
|