@voqalize/avatar 0.2.1 → 0.3.0
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/LICENSE +21 -661
- package/README.md +144 -663
- package/client/Avatar.tsx +57 -0
- package/client/AvatarClient.ts +523 -0
- package/client/arjun.ts +24 -0
- package/client/createAvatar.ts +139 -0
- package/client/createCanvasAvatar.ts +72 -0
- package/client/index.ts +39 -0
- package/client/internal.ts +80 -0
- package/client/interviewer-female.ts +4 -0
- package/client/interviewer-male.ts +4 -0
- package/client/ishita.ts +24 -0
- package/client/kabir.ts +24 -0
- package/client/meera.ts +24 -0
- package/client/naina.ts +24 -0
- package/client/professional-female-a.ts +4 -0
- package/client/professional-female-b.ts +4 -0
- package/client/professional-male-a.ts +4 -0
- package/client/professional-male-b.ts +4 -0
- package/client/react.ts +13 -0
- package/client/types.ts +146 -0
- package/client/useAvatar.ts +74 -0
- package/client/vikram.ts +24 -0
- package/dist/Avatar.d.ts +44 -0
- package/dist/Avatar.d.ts.map +1 -0
- package/{client/dist → dist}/Avatar.js +2 -2
- package/dist/Avatar.js.map +1 -0
- package/dist/AvatarClient.d.ts +196 -0
- package/dist/AvatarClient.d.ts.map +1 -0
- package/dist/AvatarClient.js +450 -0
- package/dist/AvatarClient.js.map +1 -0
- package/dist/arjun.d.ts +6 -0
- package/dist/arjun.d.ts.map +1 -0
- package/dist/arjun.js +18 -0
- package/dist/arjun.js.map +1 -0
- package/dist/createAvatar.d.ts +99 -0
- package/dist/createAvatar.d.ts.map +1 -0
- package/dist/createAvatar.js +75 -0
- package/dist/createAvatar.js.map +1 -0
- package/dist/createCanvasAvatar.d.ts +22 -0
- package/dist/createCanvasAvatar.d.ts.map +1 -0
- package/dist/createCanvasAvatar.js +47 -0
- package/dist/createCanvasAvatar.js.map +1 -0
- package/dist/index.d.ts +30 -0
- package/dist/index.d.ts.map +1 -0
- package/dist/index.js +29 -0
- package/dist/index.js.map +1 -0
- package/dist/internal.d.ts +30 -0
- package/dist/internal.d.ts.map +1 -0
- package/dist/internal.js +33 -0
- package/dist/internal.js.map +1 -0
- package/dist/interviewer-female.d.ts +4 -0
- package/dist/interviewer-female.d.ts.map +1 -0
- package/dist/interviewer-female.js +3 -0
- package/dist/interviewer-female.js.map +1 -0
- package/dist/interviewer-male.d.ts +4 -0
- package/dist/interviewer-male.d.ts.map +1 -0
- package/dist/interviewer-male.js +3 -0
- package/dist/interviewer-male.js.map +1 -0
- package/dist/ishita.d.ts +6 -0
- package/dist/ishita.d.ts.map +1 -0
- package/dist/ishita.js +18 -0
- package/dist/ishita.js.map +1 -0
- package/dist/kabir.d.ts +6 -0
- package/dist/kabir.d.ts.map +1 -0
- package/dist/kabir.js +18 -0
- package/dist/kabir.js.map +1 -0
- package/dist/meera.d.ts +6 -0
- package/dist/meera.d.ts.map +1 -0
- package/dist/meera.js +18 -0
- package/dist/meera.js.map +1 -0
- package/dist/naina.d.ts +6 -0
- package/dist/naina.d.ts.map +1 -0
- package/dist/naina.js +18 -0
- package/dist/naina.js.map +1 -0
- package/dist/professional-female-a.d.ts +4 -0
- package/dist/professional-female-a.d.ts.map +1 -0
- package/dist/professional-female-a.js +3 -0
- package/dist/professional-female-a.js.map +1 -0
- package/dist/professional-female-b.d.ts +4 -0
- package/dist/professional-female-b.d.ts.map +1 -0
- package/dist/professional-female-b.js +3 -0
- package/dist/professional-female-b.js.map +1 -0
- package/dist/professional-male-a.d.ts +4 -0
- package/dist/professional-male-a.d.ts.map +1 -0
- package/dist/professional-male-a.js +3 -0
- package/dist/professional-male-a.js.map +1 -0
- package/dist/professional-male-b.d.ts +4 -0
- package/dist/professional-male-b.d.ts.map +1 -0
- package/dist/professional-male-b.js +3 -0
- package/dist/professional-male-b.js.map +1 -0
- package/dist/react.d.ts +13 -0
- package/dist/react.d.ts.map +1 -0
- package/dist/react.js +13 -0
- package/dist/react.js.map +1 -0
- package/dist/types.d.ts +94 -0
- package/dist/types.d.ts.map +1 -0
- package/dist/types.js +93 -0
- package/dist/types.js.map +1 -0
- package/dist/useAvatar.d.ts +39 -0
- package/dist/useAvatar.d.ts.map +1 -0
- package/dist/useAvatar.js +35 -0
- package/dist/useAvatar.js.map +1 -0
- package/dist/vikram.d.ts +6 -0
- package/dist/vikram.d.ts.map +1 -0
- package/dist/vikram.js +18 -0
- package/dist/vikram.js.map +1 -0
- package/package.json +81 -24
- package/src/avatar.d.ts +161 -120
- package/src/avatar.js +234 -146
- package/src/behavior.d.ts +26 -0
- package/src/behavior.js +102 -0
- package/src/camera.js +29 -0
- package/src/canvas/author/parts/eye.mjs +722 -0
- package/src/canvas/author/parts/hand.mjs +1156 -0
- package/src/canvas/author/parts/mouth.mjs +741 -0
- package/src/canvas/author/parts/nose.mjs +100 -0
- package/src/canvas/author/parts/skin-detail.mjs +67 -0
- package/src/canvas/author/path.mjs +283 -0
- package/src/canvas/author/rig.mjs +405 -0
- package/src/canvas/avatars/round/face.d.mts +3 -0
- package/src/canvas/avatars/round/face.mjs +1307 -0
- package/src/canvas/create-rig.d.ts +15 -0
- package/src/canvas/create-rig.js +100 -0
- package/src/canvas/data/img/professional-female-a-hair-back.webp +0 -0
- package/src/canvas/data/img/professional-female-a-hair-front.webp +0 -0
- package/src/canvas/data/img/professional-female-a-top-body.webp +0 -0
- package/src/canvas/data/img/professional-female-b-hair-back.webp +0 -0
- package/src/canvas/data/img/professional-female-b-hair-front.webp +0 -0
- package/src/canvas/data/img/professional-female-b-top-body.webp +0 -0
- package/src/canvas/data/img/professional-male-a-hair-back.webp +0 -0
- package/src/canvas/data/img/professional-male-a-hair-front.webp +0 -0
- package/src/canvas/data/img/professional-male-a-top-body.webp +0 -0
- package/src/canvas/data/img/professional-male-b-hair-back.webp +0 -0
- package/src/canvas/data/img/professional-male-b-hair-front.webp +0 -0
- package/src/canvas/data/img/professional-male-b-top-body.webp +0 -0
- package/src/canvas/data/img/round-m3-hair-back.webp +0 -0
- package/src/canvas/data/img/round-m3-hair-front.webp +0 -0
- package/src/canvas/data/img/round-m3-top-body.webp +0 -0
- package/src/canvas/data/img/round-w1-hair-back.webp +0 -0
- package/src/canvas/data/img/round-w1-hair-front.webp +0 -0
- package/src/canvas/data/img/round-w1-top-body.webp +0 -0
- package/src/canvas/data/interviewer-female.rig.json +1 -0
- package/src/canvas/data/interviewer-male.rig.json +1 -0
- package/src/canvas/data/professional-female-a.rig.json +1 -0
- package/src/canvas/data/professional-female-b.rig.json +1 -0
- package/src/canvas/data/professional-male-a.rig.json +1 -0
- package/src/canvas/data/professional-male-b.rig.json +1 -0
- package/src/canvas/src/live.js +508 -0
- package/src/canvas/src/render2d.js +218 -0
- package/src/canvas/src/rig.js +297 -0
- package/src/canvas/src/vocab.js +96 -0
- package/src/clips.js +18 -7
- package/src/conformance.js +119 -0
- package/src/emotions.js +4 -3
- package/src/face-core.js +27 -1
- package/src/face-myna.d.ts +7 -0
- package/src/face-myna.js +192 -131
- package/src/face-peep-control-plane.js +167 -0
- package/src/face-peep.d.ts +7 -0
- package/src/face-peep.js +274 -116
- package/src/face-wren.d.ts +7 -0
- package/src/face-wren.js +17 -19
- package/src/faces.d.ts +12 -0
- package/src/faces.js +53 -0
- package/src/gaze.js +119 -4
- package/src/hand.js +120 -91
- package/src/idle.js +18 -101
- package/src/interjections.js +156 -31
- package/src/params.js +8 -4
- package/src/perform.js +7 -9
- package/src/rig.d.ts +24 -0
- package/src/rig.js +32 -0
- package/src/speech-timing.js +23 -0
- package/src/visemes.js +28 -12
- package/client/dist/Avatar.d.ts +0 -27
- package/client/dist/Avatar.d.ts.map +0 -1
- package/client/dist/Avatar.js.map +0 -1
- package/client/dist/AvatarClient.d.ts +0 -139
- package/client/dist/AvatarClient.d.ts.map +0 -1
- package/client/dist/AvatarClient.js +0 -223
- package/client/dist/AvatarClient.js.map +0 -1
- package/client/dist/index.d.ts +0 -22
- package/client/dist/index.d.ts.map +0 -1
- package/client/dist/index.js +0 -22
- package/client/dist/index.js.map +0 -1
- package/client/dist/types.d.ts +0 -86
- package/client/dist/types.d.ts.map +0 -1
- package/client/dist/types.js +0 -31
- package/client/dist/types.js.map +0 -1
- package/client/dist/useAvatar.d.ts +0 -42
- package/client/dist/useAvatar.d.ts.map +0 -1
- package/client/dist/useAvatar.js +0 -51
- package/client/dist/useAvatar.js.map +0 -1
- package/client/src/Avatar.tsx +0 -33
- package/client/src/AvatarClient.ts +0 -270
- package/client/src/index.ts +0 -22
- package/client/src/types.ts +0 -104
- package/client/src/useAvatar.ts +0 -85
- package/docs/contract-avatar.md +0 -371
- package/docs/contract-protocol.md +0 -461
|
@@ -1,139 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* AvatarClient — the avatar's server-message dispatcher, turn clock, and cue
|
|
3
|
-
* splice, framework-free (no React; the hook and component wrap this).
|
|
4
|
-
*
|
|
5
|
-
* ## Turn clock anchoring
|
|
6
|
-
*
|
|
7
|
-
* A turn's `t0` is anchored to `performance.now()` **at the moment this client
|
|
8
|
-
* receives the `{cmd:"speech", event:"start"}` message** — cues are
|
|
9
|
-
* client-anchored, not per-cue server-released. That message rides the RTVI
|
|
10
|
-
* data channel, ahead of the jitter-buffered audio path, so the residual error
|
|
11
|
-
* lands on the video-leads side — the side `docs/contract-protocol.md` says
|
|
12
|
-
* perceptual tolerance favours (+125 ms vs -45 ms).
|
|
13
|
-
*
|
|
14
|
-
* We investigated anchoring on pipecat client-js's own `RTVIEvent
|
|
15
|
-
* .BotStartedSpeaking`/`BotStoppedSpeaking` instead (or as a refinement) and
|
|
16
|
-
* chose not to, for two reasons:
|
|
17
|
-
*
|
|
18
|
-
* 1. **No turn correlation.** Those events carry no payload — no `ctx` — so
|
|
19
|
-
* there is no way to tell which turn a firing belongs to. Our own
|
|
20
|
-
* `speech` command carries `ctx`, which the splice logic below needs
|
|
21
|
-
* regardless, so anchoring off it costs nothing extra.
|
|
22
|
-
* 2. **Same source, same path, no accuracy gain.** The `AvatarProcessor`
|
|
23
|
-
* sits between the TTS service and the output transport and observes the
|
|
24
|
-
* transport's own `BotStarted/StoppedSpeakingFrame` broadcasts — the exact
|
|
25
|
-
* frame pipecat's built-in speaking detection is *also* driven from. Both
|
|
26
|
-
* notifications travel the same data-channel path to the browser. There is
|
|
27
|
-
* no local "truly audible now" signal cheaply available: the audio arrives
|
|
28
|
-
* on a `MediaStreamTrack` whose only lifecycle events (`unmute`/`mute`)
|
|
29
|
-
* fire once per call, not per utterance. Tapping the decoded remote audio
|
|
30
|
-
* with a WebAudio `AnalyserNode` RMS gate *would* give one, but it adds
|
|
31
|
-
* its own onset latency and a real audio pipeline to build and tune, and
|
|
32
|
-
* it would eat into the intentional video-first safety margin rather than
|
|
33
|
-
* improve it. Left as a documented option, not built.
|
|
34
|
-
*
|
|
35
|
-
* `attach()` therefore subscribes to exactly one pipecat event,
|
|
36
|
-
* `serverMessage`. It used to also subscribe to both speaking events to report
|
|
37
|
-
* a diagnostic drift between our anchor and pipecat's; that hook is gone with
|
|
38
|
-
* the rest of the observability surface (`docs/removed.md` § Client callbacks).
|
|
39
|
-
*
|
|
40
|
-
* ## Cue splice
|
|
41
|
-
*
|
|
42
|
-
* The widget has two cue-track primitives: `speak({cues, clock})` (a full
|
|
43
|
-
* replace) and `pushCues(cues)` (a pure union that can only grow the track,
|
|
44
|
-
* never shrink it). Neither is "discard queued cues at or after `from_ms`,
|
|
45
|
-
* then append" on its own — `pushCues` has no way to drop a stale tail. So
|
|
46
|
-
* this client keeps the turn's canonical cue array itself (kept portion +
|
|
47
|
-
* every appended chunk, spliced on each `cues` message) and picks the cheapest
|
|
48
|
-
* widget call that stays correct:
|
|
49
|
-
*
|
|
50
|
-
* - if the splice's `from_ms` doesn't reach back into anything already
|
|
51
|
-
* queued — the common case past a turn's first sentence, since only the
|
|
52
|
-
* first sentence genuinely plays fast-leg cues — nothing needs discarding:
|
|
53
|
-
* `pushCues(newCues)` is the cheap, correct append.
|
|
54
|
-
* - if it does reach back (a real fast→accurate splice), `pushCues` cannot
|
|
55
|
-
* express the discard; we call `speak()` again with the full spliced
|
|
56
|
-
* canonical array on the turn's original clock. `speak()` is otherwise
|
|
57
|
-
* documented as also killing an in-flight spoken interjection and
|
|
58
|
-
* re-entering `SPEAKING` — both harmless mid-splice (an interjection
|
|
59
|
-
* should not be running while a server track owns the mouth; re-entering
|
|
60
|
-
* an unchanged state is a no-op past the profile/gaze reset the widget
|
|
61
|
-
* already does for a same-name `setState`).
|
|
62
|
-
*
|
|
63
|
-
* Cues commonly arrive **before** `speech start` — the fast leg starts the
|
|
64
|
-
* moment a sentence is handed to TTS, well before `BotStartedSpeakingFrame`.
|
|
65
|
-
* Chunks that arrive before the clock is anchored are spliced into the
|
|
66
|
-
* canonical array but not yet handed to the widget; `speech start` hands over
|
|
67
|
-
* whatever has accumulated as the turn's first `speak()` call. So "the first
|
|
68
|
-
* chunk of a turn starts speak()" means the first *widget* call, not
|
|
69
|
-
* necessarily the first *message*.
|
|
70
|
-
*/
|
|
71
|
-
import type { PipecatClient } from "@pipecat-ai/client-js";
|
|
72
|
-
import type { AvatarApi } from "../../src/avatar.js";
|
|
73
|
-
import { type AvatarCommand, type AvatarCue } from "./types.js";
|
|
74
|
-
/**
|
|
75
|
-
* Internal. Not exported from the package — see `index.ts` for the public
|
|
76
|
-
* surface, which is `<Avatar>` and nothing else.
|
|
77
|
-
*
|
|
78
|
-
* There is deliberately no `accept` predicate here any more. Avatar commands
|
|
79
|
-
* travel in one envelope, `{type:"avatar"}`, in both directions and from every
|
|
80
|
-
* source: a `AvatarProcessor` in the pipeline and a brain driving the face
|
|
81
|
-
* out of band emit the same shape. A per-deployment predicate meant the
|
|
82
|
-
* library could not state what an avatar message *is*, which is the one thing
|
|
83
|
-
* a wire format has to be able to say. See docs/removed.md § The accept
|
|
84
|
-
* predicate.
|
|
85
|
-
*/
|
|
86
|
-
export interface AvatarClientOptions {
|
|
87
|
-
/** A dispatch threw (e.g. an unknown state or interjection id, which the
|
|
88
|
-
* widget throws on). Defaults to `console.warn`. */
|
|
89
|
-
onError?: (err: unknown, msg: AvatarCommand) => void;
|
|
90
|
-
/** Override for tests. Defaults to `performance.now`. */
|
|
91
|
-
now?: () => number;
|
|
92
|
-
}
|
|
93
|
-
/**
|
|
94
|
-
* The one `RTVIEvent` member `attach()` subscribes to, spelled as its value.
|
|
95
|
-
*
|
|
96
|
-
* Written out rather than imported because that enum was this module's *only*
|
|
97
|
-
* runtime reference to `@pipecat-ai/client-js`, and one runtime reference makes
|
|
98
|
-
* the whole `/pipecat` subpath fail to load without the peer installed — even
|
|
99
|
-
* for a host that drives `dispatch()` from its own transport and never calls
|
|
100
|
-
* `attach()`. The peer is declared optional; this is what makes that true
|
|
101
|
-
* rather than aspirational.
|
|
102
|
-
*
|
|
103
|
-
* String enums are nominal in TypeScript, so the compiler cannot check these
|
|
104
|
-
* against the real ones from a type-only import. `client/test/AvatarClient.test.ts`
|
|
105
|
-
* does it instead, against the actual enum — the devDependency is present
|
|
106
|
-
* exactly where the check belongs and absent from what we ship.
|
|
107
|
-
*/
|
|
108
|
-
export declare const RTVI_EVENTS: {
|
|
109
|
-
readonly serverMessage: "serverMessage";
|
|
110
|
-
};
|
|
111
|
-
export declare class AvatarClient {
|
|
112
|
-
private readonly avatar;
|
|
113
|
-
private readonly opts;
|
|
114
|
-
private readonly now;
|
|
115
|
-
private turn;
|
|
116
|
-
constructor(avatar: AvatarApi, opts?: AvatarClientOptions);
|
|
117
|
-
/** The active turn's ctx, or `null` between turns. For tests and telemetry. */
|
|
118
|
-
get turnCtx(): string | null;
|
|
119
|
-
/** The active turn's canonical (already-spliced) cue track. For tests and telemetry. */
|
|
120
|
-
get turnCues(): AvatarCue[];
|
|
121
|
-
/** Dispatch one server message. Anything that isn't in the avatar envelope
|
|
122
|
-
* is not ours and is ignored; so is an envelope carrying a `cmd` this build
|
|
123
|
-
* has never heard of, per the wire protocol's forward-compat rule. */
|
|
124
|
-
dispatch(raw: unknown): void;
|
|
125
|
-
private handleState;
|
|
126
|
-
private ensureTurn;
|
|
127
|
-
private handleCues;
|
|
128
|
-
private handleSpeech;
|
|
129
|
-
/**
|
|
130
|
-
* Subscribe to a live `PipecatClient`'s server messages and dispatch the
|
|
131
|
-
* avatar commands among them — the ones in the protocol's own
|
|
132
|
-
* `{type:"avatar"}` envelope, which `isAvatarMessage` is the definition of.
|
|
133
|
-
* Never throws on a malformed or irrelevant message.
|
|
134
|
-
*
|
|
135
|
-
* @returns an unsubscribe function; call it on unmount or disconnect.
|
|
136
|
-
*/
|
|
137
|
-
attach(client: PipecatClient): () => void;
|
|
138
|
-
}
|
|
139
|
-
//# sourceMappingURL=AvatarClient.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"AvatarClient.d.ts","sourceRoot":"","sources":["../src/AvatarClient.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqEG;AAEH,OAAO,KAAK,EAAE,aAAa,EAAa,MAAM,uBAAuB,CAAC;AACtE,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,qBAAqB,CAAC;AACrD,OAAO,EAEL,KAAK,aAAa,EAClB,KAAK,SAAS,EAIf,MAAM,YAAY,CAAC;AAYpB;;;;;;;;;;;GAWG;AACH,MAAM,WAAW,mBAAmB;IAClC;wDACoD;IACpD,OAAO,CAAC,EAAE,CAAC,GAAG,EAAE,OAAO,EAAE,GAAG,EAAE,aAAa,KAAK,IAAI,CAAC;IACrD,yDAAyD;IACzD,GAAG,CAAC,EAAE,MAAM,MAAM,CAAC;CACpB;AAED;;;;;;;;;;;;;;GAcG;AACH,eAAO,MAAM,WAAW;;CAEmB,CAAC;AAU5C,qBAAa,YAAY;IACvB,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAY;IACnC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAsB;IAC3C,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAe;IACnC,OAAO,CAAC,IAAI,CAAqB;gBAErB,MAAM,EAAE,SAAS,EAAE,IAAI,GAAE,mBAAwB;IAM7D,+EAA+E;IAC/E,IAAI,OAAO,IAAI,MAAM,GAAG,IAAI,CAE3B;IAED,wFAAwF;IACxF,IAAI,QAAQ,IAAI,SAAS,EAAE,CAE1B;IAED;;0EAEsE;IACtE,QAAQ,CAAC,GAAG,EAAE,OAAO,GAAG,IAAI;IAkC5B,OAAO,CAAC,WAAW;IAUnB,OAAO,CAAC,UAAU;IAWlB,OAAO,CAAC,UAAU;IAkBlB,OAAO,CAAC,YAAY;IAkBpB;;;;;;;OAOG;IACH,MAAM,CAAC,MAAM,EAAE,aAAa,GAAG,MAAM,IAAI;CAM1C"}
|
|
@@ -1,223 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* AvatarClient — the avatar's server-message dispatcher, turn clock, and cue
|
|
3
|
-
* splice, framework-free (no React; the hook and component wrap this).
|
|
4
|
-
*
|
|
5
|
-
* ## Turn clock anchoring
|
|
6
|
-
*
|
|
7
|
-
* A turn's `t0` is anchored to `performance.now()` **at the moment this client
|
|
8
|
-
* receives the `{cmd:"speech", event:"start"}` message** — cues are
|
|
9
|
-
* client-anchored, not per-cue server-released. That message rides the RTVI
|
|
10
|
-
* data channel, ahead of the jitter-buffered audio path, so the residual error
|
|
11
|
-
* lands on the video-leads side — the side `docs/contract-protocol.md` says
|
|
12
|
-
* perceptual tolerance favours (+125 ms vs -45 ms).
|
|
13
|
-
*
|
|
14
|
-
* We investigated anchoring on pipecat client-js's own `RTVIEvent
|
|
15
|
-
* .BotStartedSpeaking`/`BotStoppedSpeaking` instead (or as a refinement) and
|
|
16
|
-
* chose not to, for two reasons:
|
|
17
|
-
*
|
|
18
|
-
* 1. **No turn correlation.** Those events carry no payload — no `ctx` — so
|
|
19
|
-
* there is no way to tell which turn a firing belongs to. Our own
|
|
20
|
-
* `speech` command carries `ctx`, which the splice logic below needs
|
|
21
|
-
* regardless, so anchoring off it costs nothing extra.
|
|
22
|
-
* 2. **Same source, same path, no accuracy gain.** The `AvatarProcessor`
|
|
23
|
-
* sits between the TTS service and the output transport and observes the
|
|
24
|
-
* transport's own `BotStarted/StoppedSpeakingFrame` broadcasts — the exact
|
|
25
|
-
* frame pipecat's built-in speaking detection is *also* driven from. Both
|
|
26
|
-
* notifications travel the same data-channel path to the browser. There is
|
|
27
|
-
* no local "truly audible now" signal cheaply available: the audio arrives
|
|
28
|
-
* on a `MediaStreamTrack` whose only lifecycle events (`unmute`/`mute`)
|
|
29
|
-
* fire once per call, not per utterance. Tapping the decoded remote audio
|
|
30
|
-
* with a WebAudio `AnalyserNode` RMS gate *would* give one, but it adds
|
|
31
|
-
* its own onset latency and a real audio pipeline to build and tune, and
|
|
32
|
-
* it would eat into the intentional video-first safety margin rather than
|
|
33
|
-
* improve it. Left as a documented option, not built.
|
|
34
|
-
*
|
|
35
|
-
* `attach()` therefore subscribes to exactly one pipecat event,
|
|
36
|
-
* `serverMessage`. It used to also subscribe to both speaking events to report
|
|
37
|
-
* a diagnostic drift between our anchor and pipecat's; that hook is gone with
|
|
38
|
-
* the rest of the observability surface (`docs/removed.md` § Client callbacks).
|
|
39
|
-
*
|
|
40
|
-
* ## Cue splice
|
|
41
|
-
*
|
|
42
|
-
* The widget has two cue-track primitives: `speak({cues, clock})` (a full
|
|
43
|
-
* replace) and `pushCues(cues)` (a pure union that can only grow the track,
|
|
44
|
-
* never shrink it). Neither is "discard queued cues at or after `from_ms`,
|
|
45
|
-
* then append" on its own — `pushCues` has no way to drop a stale tail. So
|
|
46
|
-
* this client keeps the turn's canonical cue array itself (kept portion +
|
|
47
|
-
* every appended chunk, spliced on each `cues` message) and picks the cheapest
|
|
48
|
-
* widget call that stays correct:
|
|
49
|
-
*
|
|
50
|
-
* - if the splice's `from_ms` doesn't reach back into anything already
|
|
51
|
-
* queued — the common case past a turn's first sentence, since only the
|
|
52
|
-
* first sentence genuinely plays fast-leg cues — nothing needs discarding:
|
|
53
|
-
* `pushCues(newCues)` is the cheap, correct append.
|
|
54
|
-
* - if it does reach back (a real fast→accurate splice), `pushCues` cannot
|
|
55
|
-
* express the discard; we call `speak()` again with the full spliced
|
|
56
|
-
* canonical array on the turn's original clock. `speak()` is otherwise
|
|
57
|
-
* documented as also killing an in-flight spoken interjection and
|
|
58
|
-
* re-entering `SPEAKING` — both harmless mid-splice (an interjection
|
|
59
|
-
* should not be running while a server track owns the mouth; re-entering
|
|
60
|
-
* an unchanged state is a no-op past the profile/gaze reset the widget
|
|
61
|
-
* already does for a same-name `setState`).
|
|
62
|
-
*
|
|
63
|
-
* Cues commonly arrive **before** `speech start` — the fast leg starts the
|
|
64
|
-
* moment a sentence is handed to TTS, well before `BotStartedSpeakingFrame`.
|
|
65
|
-
* Chunks that arrive before the clock is anchored are spliced into the
|
|
66
|
-
* canonical array but not yet handed to the widget; `speech start` hands over
|
|
67
|
-
* whatever has accumulated as the turn's first `speak()` call. So "the first
|
|
68
|
-
* chunk of a turn starts speak()" means the first *widget* call, not
|
|
69
|
-
* necessarily the first *message*.
|
|
70
|
-
*/
|
|
71
|
-
import { isAvatarMessage, } from "./types.js";
|
|
72
|
-
/**
|
|
73
|
-
* The one `RTVIEvent` member `attach()` subscribes to, spelled as its value.
|
|
74
|
-
*
|
|
75
|
-
* Written out rather than imported because that enum was this module's *only*
|
|
76
|
-
* runtime reference to `@pipecat-ai/client-js`, and one runtime reference makes
|
|
77
|
-
* the whole `/pipecat` subpath fail to load without the peer installed — even
|
|
78
|
-
* for a host that drives `dispatch()` from its own transport and never calls
|
|
79
|
-
* `attach()`. The peer is declared optional; this is what makes that true
|
|
80
|
-
* rather than aspirational.
|
|
81
|
-
*
|
|
82
|
-
* String enums are nominal in TypeScript, so the compiler cannot check these
|
|
83
|
-
* against the real ones from a type-only import. `client/test/AvatarClient.test.ts`
|
|
84
|
-
* does it instead, against the actual enum — the devDependency is present
|
|
85
|
-
* exactly where the check belongs and absent from what we ship.
|
|
86
|
-
*/
|
|
87
|
-
export const RTVI_EVENTS = {
|
|
88
|
-
serverMessage: "serverMessage",
|
|
89
|
-
};
|
|
90
|
-
/** Defensive unwrap for the `RTVIEvent.ServerMessage` `{ data }` quirk: some
|
|
91
|
-
* transports deliver the payload directly and some wrap it once more. */
|
|
92
|
-
function unwrapServerMessage(raw) {
|
|
93
|
-
const obj = (raw ?? {});
|
|
94
|
-
const inner = obj["data"];
|
|
95
|
-
return inner && "type" in inner ? inner : obj;
|
|
96
|
-
}
|
|
97
|
-
export class AvatarClient {
|
|
98
|
-
avatar;
|
|
99
|
-
opts;
|
|
100
|
-
now;
|
|
101
|
-
turn = null;
|
|
102
|
-
constructor(avatar, opts = {}) {
|
|
103
|
-
this.avatar = avatar;
|
|
104
|
-
this.opts = opts;
|
|
105
|
-
this.now = opts.now ?? (() => performance.now());
|
|
106
|
-
}
|
|
107
|
-
/** The active turn's ctx, or `null` between turns. For tests and telemetry. */
|
|
108
|
-
get turnCtx() {
|
|
109
|
-
return this.turn?.ctx ?? null;
|
|
110
|
-
}
|
|
111
|
-
/** The active turn's canonical (already-spliced) cue track. For tests and telemetry. */
|
|
112
|
-
get turnCues() {
|
|
113
|
-
return this.turn ? [...this.turn.cues] : [];
|
|
114
|
-
}
|
|
115
|
-
/** Dispatch one server message. Anything that isn't in the avatar envelope
|
|
116
|
-
* is not ours and is ignored; so is an envelope carrying a `cmd` this build
|
|
117
|
-
* has never heard of, per the wire protocol's forward-compat rule. */
|
|
118
|
-
dispatch(raw) {
|
|
119
|
-
if (!isAvatarMessage(raw))
|
|
120
|
-
return;
|
|
121
|
-
const msg = raw;
|
|
122
|
-
try {
|
|
123
|
-
switch (msg.cmd) {
|
|
124
|
-
case "state":
|
|
125
|
-
this.handleState(msg);
|
|
126
|
-
break;
|
|
127
|
-
case "interject":
|
|
128
|
-
this.avatar.interject(msg.id);
|
|
129
|
-
break;
|
|
130
|
-
case "gesture":
|
|
131
|
-
this.avatar.gesture(msg.id);
|
|
132
|
-
break;
|
|
133
|
-
case "cues":
|
|
134
|
-
this.handleCues(msg);
|
|
135
|
-
break;
|
|
136
|
-
case "speech":
|
|
137
|
-
this.handleSpeech(msg);
|
|
138
|
-
break;
|
|
139
|
-
case "user":
|
|
140
|
-
this.avatar.setUserSpeaking(msg.speaking);
|
|
141
|
-
break;
|
|
142
|
-
// No default: an unknown `cmd` is a newer server talking to an older
|
|
143
|
-
// widget, and the protocol's forward-compat rule says ignore it. There
|
|
144
|
-
// is no callback for it — a hook nobody could act on is observability,
|
|
145
|
-
// not an interface (`docs/removed.md` § Client callbacks).
|
|
146
|
-
}
|
|
147
|
-
}
|
|
148
|
-
catch (err) {
|
|
149
|
-
if (this.opts.onError)
|
|
150
|
-
this.opts.onError(err, msg);
|
|
151
|
-
else
|
|
152
|
-
console.warn("[avatar] dispatch failed", msg, err);
|
|
153
|
-
}
|
|
154
|
-
}
|
|
155
|
-
handleState(msg) {
|
|
156
|
-
// Deliberately no client-side dedup: pass every `state` command straight
|
|
157
|
-
// through. The widget's own setState already no-ops the parts that matter
|
|
158
|
-
// for an unchanged name (`changed` gates the blink and the 'state' event in
|
|
159
|
-
// avatar.js), and a server resending the same state name as a
|
|
160
|
-
// keepalive/resync must still land so an `emotion`/`gaze` override on this
|
|
161
|
-
// particular message takes effect.
|
|
162
|
-
this.avatar.setState(msg.name, { emotion: msg.emotion, gaze: msg.gaze });
|
|
163
|
-
}
|
|
164
|
-
ensureTurn(ctx) {
|
|
165
|
-
if (!this.turn || this.turn.ctx !== ctx) {
|
|
166
|
-
// A different ctx supersedes whatever turn we had — a stale trailing
|
|
167
|
-
// message for the old ctx will find `this.turn.ctx !== ctx` in
|
|
168
|
-
// handleSpeech's stop-guard and be ignored, rather than cutting off the
|
|
169
|
-
// new turn.
|
|
170
|
-
this.turn = { ctx, cues: [], started: false, clock: null };
|
|
171
|
-
}
|
|
172
|
-
return this.turn;
|
|
173
|
-
}
|
|
174
|
-
handleCues(msg) {
|
|
175
|
-
const turn = this.ensureTurn(msg.ctx);
|
|
176
|
-
const kept = turn.cues.filter((c) => c.t < msg.from_ms);
|
|
177
|
-
const discarded = turn.cues.length - kept.length;
|
|
178
|
-
turn.cues = [...kept, ...msg.cues].sort((a, b) => a.t - b.t);
|
|
179
|
-
if (!turn.started) {
|
|
180
|
-
// No clock yet — buffer. `speech start` will hand this over as the turn's
|
|
181
|
-
// first speak() call.
|
|
182
|
-
return;
|
|
183
|
-
}
|
|
184
|
-
if (discarded === 0) {
|
|
185
|
-
this.avatar.pushCues(msg.cues);
|
|
186
|
-
}
|
|
187
|
-
else {
|
|
188
|
-
this.avatar.speak({ cues: turn.cues, clock: turn.clock });
|
|
189
|
-
}
|
|
190
|
-
}
|
|
191
|
-
handleSpeech(msg) {
|
|
192
|
-
if (msg.event === "start") {
|
|
193
|
-
const turn = this.ensureTurn(msg.ctx);
|
|
194
|
-
const t0 = this.now();
|
|
195
|
-
const clock = () => this.now() - t0;
|
|
196
|
-
turn.clock = clock;
|
|
197
|
-
turn.started = true;
|
|
198
|
-
this.avatar.speak({ cues: turn.cues, clock });
|
|
199
|
-
return;
|
|
200
|
-
}
|
|
201
|
-
// "stop": only act if it names the turn we're actually riding. A stale stop
|
|
202
|
-
// for an already-superseded ctx must not cut off a newer turn.
|
|
203
|
-
if (this.turn && this.turn.ctx === msg.ctx) {
|
|
204
|
-
this.avatar.stopSpeaking();
|
|
205
|
-
this.turn = null;
|
|
206
|
-
}
|
|
207
|
-
}
|
|
208
|
-
/**
|
|
209
|
-
* Subscribe to a live `PipecatClient`'s server messages and dispatch the
|
|
210
|
-
* avatar commands among them — the ones in the protocol's own
|
|
211
|
-
* `{type:"avatar"}` envelope, which `isAvatarMessage` is the definition of.
|
|
212
|
-
* Never throws on a malformed or irrelevant message.
|
|
213
|
-
*
|
|
214
|
-
* @returns an unsubscribe function; call it on unmount or disconnect.
|
|
215
|
-
*/
|
|
216
|
-
attach(client) {
|
|
217
|
-
const onServerMessage = (raw) => this.dispatch(unwrapServerMessage(raw));
|
|
218
|
-
const serverMessage = RTVI_EVENTS.serverMessage;
|
|
219
|
-
client.on(serverMessage, onServerMessage);
|
|
220
|
-
return () => client.off(serverMessage, onServerMessage);
|
|
221
|
-
}
|
|
222
|
-
}
|
|
223
|
-
//# sourceMappingURL=AvatarClient.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"AvatarClient.js","sourceRoot":"","sources":["../src/AvatarClient.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqEG;AAIH,OAAO,EACL,eAAe,GAMhB,MAAM,YAAY,CAAC;AAgCpB;;;;;;;;;;;;;;GAcG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG;IACzB,aAAa,EAAE,eAAe;CACW,CAAC;AAE5C;yEACyE;AACzE,SAAS,mBAAmB,CAAC,GAAY;IACvC,MAAM,GAAG,GAAG,CAAC,GAAG,IAAI,EAAE,CAA4B,CAAC;IACnD,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAwC,CAAC;IACjE,OAAO,KAAK,IAAI,MAAM,IAAI,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC;AAChD,CAAC;AAED,MAAM,OAAO,YAAY;IACN,MAAM,CAAY;IAClB,IAAI,CAAsB;IAC1B,GAAG,CAAe;IAC3B,IAAI,GAAgB,IAAI,CAAC;IAEjC,YAAY,MAAiB,EAAE,OAA4B,EAAE;QAC3D,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC;QACjB,IAAI,CAAC,GAAG,GAAG,IAAI,CAAC,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC,WAAW,CAAC,GAAG,EAAE,CAAC,CAAC;IACnD,CAAC;IAED,+EAA+E;IAC/E,IAAI,OAAO;QACT,OAAO,IAAI,CAAC,IAAI,EAAE,GAAG,IAAI,IAAI,CAAC;IAChC,CAAC;IAED,wFAAwF;IACxF,IAAI,QAAQ;QACV,OAAO,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;IAC9C,CAAC;IAED;;0EAEsE;IACtE,QAAQ,CAAC,GAAY;QACnB,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC;YAAE,OAAO;QAClC,MAAM,GAAG,GAAkB,GAAG,CAAC;QAC/B,IAAI,CAAC;YACH,QAAQ,GAAG,CAAC,GAAG,EAAE,CAAC;gBAChB,KAAK,OAAO;oBACV,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;oBACtB,MAAM;gBACR,KAAK,WAAW;oBACd,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;oBAC9B,MAAM;gBACR,KAAK,SAAS;oBACZ,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;oBAC5B,MAAM;gBACR,KAAK,MAAM;oBACT,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC;oBACrB,MAAM;gBACR,KAAK,QAAQ;oBACX,IAAI,CAAC,YAAY,CAAC,GAAG,CAAC,CAAC;oBACvB,MAAM;gBACR,KAAK,MAAM;oBACT,IAAI,CAAC,MAAM,CAAC,eAAe,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;oBAC1C,MAAM;gBACR,qEAAqE;gBACrE,uEAAuE;gBACvE,uEAAuE;gBACvE,2DAA2D;YAC7D,CAAC;QACH,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,IAAI,IAAI,CAAC,IAAI,CAAC,OAAO;gBAAE,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;;gBAC9C,OAAO,CAAC,IAAI,CAAC,0BAA0B,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;QAC1D,CAAC;IACH,CAAC;IAEO,WAAW,CAAC,GAAmB;QACrC,yEAAyE;QACzE,0EAA0E;QAC1E,4EAA4E;QAC5E,8DAA8D;QAC9D,2EAA2E;QAC3E,mCAAmC;QACnC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,EAAE,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC;IAC3E,CAAC;IAEO,UAAU,CAAC,GAAW;QAC5B,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,GAAG,KAAK,GAAG,EAAE,CAAC;YACxC,qEAAqE;YACrE,+DAA+D;YAC/D,wEAAwE;YACxE,YAAY;YACZ,IAAI,CAAC,IAAI,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,EAAE,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC;QAC7D,CAAC;QACD,OAAO,IAAI,CAAC,IAAI,CAAC;IACnB,CAAC;IAEO,UAAU,CAAC,GAAkB;QACnC,MAAM,IAAI,GAAG,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QACtC,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,OAAO,CAAC,CAAC;QACxD,MAAM,SAAS,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC;QACjD,IAAI,CAAC,IAAI,GAAG,CAAC,GAAG,IAAI,EAAE,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QAE7D,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC;YAClB,0EAA0E;YAC1E,sBAAsB;YACtB,OAAO;QACT,CAAC;QACD,IAAI,SAAS,KAAK,CAAC,EAAE,CAAC;YACpB,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACjC,CAAC;aAAM,CAAC;YACN,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,IAAI,CAAC,KAAM,EAAE,CAAC,CAAC;QAC7D,CAAC;IACH,CAAC;IAEO,YAAY,CAAC,GAAoB;QACvC,IAAI,GAAG,CAAC,KAAK,KAAK,OAAO,EAAE,CAAC;YAC1B,MAAM,IAAI,GAAG,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACtC,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;YACtB,MAAM,KAAK,GAAG,GAAG,EAAE,CAAC,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC;YACpC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;YACnB,IAAI,CAAC,OAAO,GAAG,IAAI,CAAC;YACpB,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;YAC9C,OAAO;QACT,CAAC;QACD,4EAA4E;QAC5E,+DAA+D;QAC/D,IAAI,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,GAAG,KAAK,GAAG,CAAC,GAAG,EAAE,CAAC;YAC3C,IAAI,CAAC,MAAM,CAAC,YAAY,EAAE,CAAC;YAC3B,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC;QACnB,CAAC;IACH,CAAC;IAED;;;;;;;OAOG;IACH,MAAM,CAAC,MAAqB;QAC1B,MAAM,eAAe,GAAG,CAAC,GAAY,EAAE,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,mBAAmB,CAAC,GAAG,CAAC,CAAC,CAAC;QAClF,MAAM,aAAa,GAAG,WAAW,CAAC,aAA0B,CAAC;QAC7D,MAAM,CAAC,EAAE,CAAC,aAAa,EAAE,eAAe,CAAC,CAAC;QAC1C,OAAO,GAAG,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,aAAa,EAAE,eAAe,CAAC,CAAC;IAC1D,CAAC;CACF"}
|
package/client/dist/index.d.ts
DELETED
|
@@ -1,22 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* `@voqalize/avatar` — the avatar as one React component.
|
|
3
|
-
*
|
|
4
|
-
* import { Avatar } from "@voqalize/avatar";
|
|
5
|
-
*
|
|
6
|
-
* <Avatar client={pipecatClient} className="avatar-tile" />
|
|
7
|
-
*
|
|
8
|
-
* That is the whole public surface. Put an `AvatarProcessor` in the pipecat
|
|
9
|
-
* pipeline (`pip install voqalize-avatar`), drop this component into the bot's
|
|
10
|
-
* tile, and the face listens, thinks, claims the floor and lipsyncs what the
|
|
11
|
-
* TTS says.
|
|
12
|
-
*
|
|
13
|
-
* The widget underneath is framework-free, and the dispatcher between it and
|
|
14
|
-
* the RTVI data channel is plain TypeScript — but neither is exported. Two
|
|
15
|
-
* consumers wanted a call tile, both are React, and a public API is a promise
|
|
16
|
-
* we have to keep across versions. `docs/removed.md` lists what used to be
|
|
17
|
-
* here and how to get it back if a real third case argues for it.
|
|
18
|
-
*
|
|
19
|
-
* Peers: `react >= 18` and `@pipecat-ai/client-js`.
|
|
20
|
-
*/
|
|
21
|
-
export { Avatar, type AvatarProps } from "./Avatar.js";
|
|
22
|
-
//# sourceMappingURL=index.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAEH,OAAO,EAAE,MAAM,EAAE,KAAK,WAAW,EAAE,MAAM,aAAa,CAAC"}
|
package/client/dist/index.js
DELETED
|
@@ -1,22 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* `@voqalize/avatar` — the avatar as one React component.
|
|
3
|
-
*
|
|
4
|
-
* import { Avatar } from "@voqalize/avatar";
|
|
5
|
-
*
|
|
6
|
-
* <Avatar client={pipecatClient} className="avatar-tile" />
|
|
7
|
-
*
|
|
8
|
-
* That is the whole public surface. Put an `AvatarProcessor` in the pipecat
|
|
9
|
-
* pipeline (`pip install voqalize-avatar`), drop this component into the bot's
|
|
10
|
-
* tile, and the face listens, thinks, claims the floor and lipsyncs what the
|
|
11
|
-
* TTS says.
|
|
12
|
-
*
|
|
13
|
-
* The widget underneath is framework-free, and the dispatcher between it and
|
|
14
|
-
* the RTVI data channel is plain TypeScript — but neither is exported. Two
|
|
15
|
-
* consumers wanted a call tile, both are React, and a public API is a promise
|
|
16
|
-
* we have to keep across versions. `docs/removed.md` lists what used to be
|
|
17
|
-
* here and how to get it back if a real third case argues for it.
|
|
18
|
-
*
|
|
19
|
-
* Peers: `react >= 18` and `@pipecat-ai/client-js`.
|
|
20
|
-
*/
|
|
21
|
-
export { Avatar } from "./Avatar.js";
|
|
22
|
-
//# sourceMappingURL=index.js.map
|
package/client/dist/index.js.map
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAEH,OAAO,EAAE,MAAM,EAAoB,MAAM,aAAa,CAAC"}
|
package/client/dist/types.d.ts
DELETED
|
@@ -1,86 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* types.ts — the avatar wire vocabulary, client side.
|
|
3
|
-
*
|
|
4
|
-
* The binding definition is `docs/contract-protocol.md`; this file is its
|
|
5
|
-
* TypeScript restatement and must not drift from it. The Python half of the
|
|
6
|
-
* same vocabulary is `py/src/voqalize_avatar/messages.py` — the three are
|
|
7
|
-
* maintained together, and a command added to one without the others is
|
|
8
|
-
* incomplete.
|
|
9
|
-
*
|
|
10
|
-
* A server pushes these as RTVI `server-message`s under the envelope
|
|
11
|
-
* `{ type: "avatar", ...cmd-specific fields }`. {@link AvatarCommand} describes
|
|
12
|
-
* what rides inside that envelope; {@link isAvatarMessage} is the envelope
|
|
13
|
-
* itself, and is the only definition of "this message is for the avatar" the
|
|
14
|
-
* client has. There is no protocol version field — see `docs/removed.md`
|
|
15
|
-
* § The `v` field.
|
|
16
|
-
*/
|
|
17
|
-
/** A viseme cue: `t` is a ms offset into the utterance's clock, `v` is a Rhubarb A–H (or X) letter. */
|
|
18
|
-
export interface AvatarCue {
|
|
19
|
-
t: number;
|
|
20
|
-
v: string;
|
|
21
|
-
i?: number;
|
|
22
|
-
}
|
|
23
|
-
export interface AvatarStateCmd {
|
|
24
|
-
cmd: "state";
|
|
25
|
-
name: string;
|
|
26
|
-
emotion?: string;
|
|
27
|
-
gaze?: string;
|
|
28
|
-
}
|
|
29
|
-
export interface AvatarInterjectCmd {
|
|
30
|
-
cmd: "interject";
|
|
31
|
-
id: string;
|
|
32
|
-
}
|
|
33
|
-
/**
|
|
34
|
-
* A hand gesture — the hand at the frame edge plus its face half. Separate from
|
|
35
|
-
* `interject` on purpose: `interject("WAVE")` is the face alone and always was,
|
|
36
|
-
* so a server that upgrades gets no hand until it asks for one.
|
|
37
|
-
*/
|
|
38
|
-
export interface AvatarGestureCmd {
|
|
39
|
-
cmd: "gesture";
|
|
40
|
-
id: string;
|
|
41
|
-
}
|
|
42
|
-
export interface AvatarCuesCmd {
|
|
43
|
-
cmd: "cues";
|
|
44
|
-
ctx: string;
|
|
45
|
-
/** Discard queued cues at or after this offset (ms), then append `cues`. */
|
|
46
|
-
from_ms: number;
|
|
47
|
-
cues: AvatarCue[];
|
|
48
|
-
/**
|
|
49
|
-
* True on the one chunk that completes this turn's track: the TTS context is
|
|
50
|
-
* closed, so no further chunk will splice into `ctx`. What a client may
|
|
51
|
-
* assume, exactly — nothing about playout. The audio it describes is still
|
|
52
|
-
* ahead, and `speech stop` remains the end of the turn. It is safe to release
|
|
53
|
-
* per-turn cue state (the splice buffer for `ctx`) once the last cue has
|
|
54
|
-
* played, and safe to stop expecting more.
|
|
55
|
-
*
|
|
56
|
-
* Absent on an interrupted turn, deliberately: a turn that was cut never
|
|
57
|
-
* claims to have completed. Absent chunks are the normal case — the widget's
|
|
58
|
-
* own track already completes on the trailing `X`, so ignoring `final`
|
|
59
|
-
* entirely is a correct implementation.
|
|
60
|
-
*/
|
|
61
|
-
final?: boolean;
|
|
62
|
-
}
|
|
63
|
-
export interface AvatarSpeechCmd {
|
|
64
|
-
cmd: "speech";
|
|
65
|
-
event: "start" | "stop";
|
|
66
|
-
ctx: string;
|
|
67
|
-
}
|
|
68
|
-
export interface AvatarUserCmd {
|
|
69
|
-
cmd: "user";
|
|
70
|
-
speaking: boolean;
|
|
71
|
-
}
|
|
72
|
-
export type AvatarCommand = AvatarStateCmd | AvatarInterjectCmd | AvatarGestureCmd | AvatarCuesCmd | AvatarSpeechCmd | AvatarUserCmd;
|
|
73
|
-
/** The full server-message payload: the envelope plus its command. */
|
|
74
|
-
export type AvatarServerMessage = AvatarCommand & {
|
|
75
|
-
type: "avatar";
|
|
76
|
-
};
|
|
77
|
-
/**
|
|
78
|
-
* Is this server-message payload the avatar's? The envelope is the whole
|
|
79
|
-
* answer: `{type:"avatar"}` with a string `cmd`. It used to be a per-deployment
|
|
80
|
-
* `accept` predicate on the client, which meant the library could not state
|
|
81
|
-
* what an avatar message *is* — see `docs/removed.md` § The accept predicate.
|
|
82
|
-
*/
|
|
83
|
-
export declare function isAvatarMessage(msg: unknown): msg is AvatarServerMessage;
|
|
84
|
-
/** The envelope `type` the protocol reserves for avatar traffic. */
|
|
85
|
-
export declare const AVATAR_MESSAGE_TYPE = "avatar";
|
|
86
|
-
//# sourceMappingURL=types.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAEH,uGAAuG;AACvG,MAAM,WAAW,SAAS;IACxB,CAAC,EAAE,MAAM,CAAC;IACV,CAAC,EAAE,MAAM,CAAC;IACV,CAAC,CAAC,EAAE,MAAM,CAAC;CACZ;AAED,MAAM,WAAW,cAAc;IAC7B,GAAG,EAAE,OAAO,CAAC;IACb,IAAI,EAAE,MAAM,CAAC;IACb,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;CACf;AAED,MAAM,WAAW,kBAAkB;IACjC,GAAG,EAAE,WAAW,CAAC;IACjB,EAAE,EAAE,MAAM,CAAC;CACZ;AAED;;;;GAIG;AACH,MAAM,WAAW,gBAAgB;IAC/B,GAAG,EAAE,SAAS,CAAC;IACf,EAAE,EAAE,MAAM,CAAC;CACZ;AAED,MAAM,WAAW,aAAa;IAC5B,GAAG,EAAE,MAAM,CAAC;IACZ,GAAG,EAAE,MAAM,CAAC;IACZ,4EAA4E;IAC5E,OAAO,EAAE,MAAM,CAAC;IAChB,IAAI,EAAE,SAAS,EAAE,CAAC;IAClB;;;;;;;;;;;;OAYG;IACH,KAAK,CAAC,EAAE,OAAO,CAAC;CACjB;AAED,MAAM,WAAW,eAAe;IAC9B,GAAG,EAAE,QAAQ,CAAC;IACd,KAAK,EAAE,OAAO,GAAG,MAAM,CAAC;IACxB,GAAG,EAAE,MAAM,CAAC;CACb;AAED,MAAM,WAAW,aAAa;IAC5B,GAAG,EAAE,MAAM,CAAC;IACZ,QAAQ,EAAE,OAAO,CAAC;CACnB;AAED,MAAM,MAAM,aAAa,GACrB,cAAc,GACd,kBAAkB,GAClB,gBAAgB,GAChB,aAAa,GACb,eAAe,GACf,aAAa,CAAC;AAElB,sEAAsE;AACtE,MAAM,MAAM,mBAAmB,GAAG,aAAa,GAAG;IAAE,IAAI,EAAE,QAAQ,CAAA;CAAE,CAAC;AAErE;;;;;GAKG;AACH,wBAAgB,eAAe,CAAC,GAAG,EAAE,OAAO,GAAG,GAAG,IAAI,mBAAmB,CAIxE;AAED,oEAAoE;AACpE,eAAO,MAAM,mBAAmB,WAAW,CAAC"}
|
package/client/dist/types.js
DELETED
|
@@ -1,31 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* types.ts — the avatar wire vocabulary, client side.
|
|
3
|
-
*
|
|
4
|
-
* The binding definition is `docs/contract-protocol.md`; this file is its
|
|
5
|
-
* TypeScript restatement and must not drift from it. The Python half of the
|
|
6
|
-
* same vocabulary is `py/src/voqalize_avatar/messages.py` — the three are
|
|
7
|
-
* maintained together, and a command added to one without the others is
|
|
8
|
-
* incomplete.
|
|
9
|
-
*
|
|
10
|
-
* A server pushes these as RTVI `server-message`s under the envelope
|
|
11
|
-
* `{ type: "avatar", ...cmd-specific fields }`. {@link AvatarCommand} describes
|
|
12
|
-
* what rides inside that envelope; {@link isAvatarMessage} is the envelope
|
|
13
|
-
* itself, and is the only definition of "this message is for the avatar" the
|
|
14
|
-
* client has. There is no protocol version field — see `docs/removed.md`
|
|
15
|
-
* § The `v` field.
|
|
16
|
-
*/
|
|
17
|
-
/**
|
|
18
|
-
* Is this server-message payload the avatar's? The envelope is the whole
|
|
19
|
-
* answer: `{type:"avatar"}` with a string `cmd`. It used to be a per-deployment
|
|
20
|
-
* `accept` predicate on the client, which meant the library could not state
|
|
21
|
-
* what an avatar message *is* — see `docs/removed.md` § The accept predicate.
|
|
22
|
-
*/
|
|
23
|
-
export function isAvatarMessage(msg) {
|
|
24
|
-
if (typeof msg !== "object" || msg === null)
|
|
25
|
-
return false;
|
|
26
|
-
const m = msg;
|
|
27
|
-
return m.type === AVATAR_MESSAGE_TYPE && typeof m.cmd === "string";
|
|
28
|
-
}
|
|
29
|
-
/** The envelope `type` the protocol reserves for avatar traffic. */
|
|
30
|
-
export const AVATAR_MESSAGE_TYPE = "avatar";
|
|
31
|
-
//# sourceMappingURL=types.js.map
|
package/client/dist/types.js.map
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"types.js","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AA2EH;;;;;GAKG;AACH,MAAM,UAAU,eAAe,CAAC,GAAY;IAC1C,IAAI,OAAO,GAAG,KAAK,QAAQ,IAAI,GAAG,KAAK,IAAI;QAAE,OAAO,KAAK,CAAC;IAC1D,MAAM,CAAC,GAAG,GAA8B,CAAC;IACzC,OAAO,CAAC,CAAC,IAAI,KAAK,mBAAmB,IAAI,OAAO,CAAC,CAAC,GAAG,KAAK,QAAQ,CAAC;AACrE,CAAC;AAED,oEAAoE;AACpE,MAAM,CAAC,MAAM,mBAAmB,GAAG,QAAQ,CAAC"}
|
|
@@ -1,42 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* useAvatar — mount the widget, wire it to a live session, dispatch its
|
|
3
|
-
* server-messages, and clean up.
|
|
4
|
-
*
|
|
5
|
-
* Internal: `<Avatar>` is the only thing the package exports. It is a separate
|
|
6
|
-
* module anyway because the two lifecycles genuinely differ — mounting the
|
|
7
|
-
* widget happens once (an avatar swap remounts by design; see the note in the
|
|
8
|
-
* effect), while attaching to the pipecat client re-runs whenever the client
|
|
9
|
-
* identity changes, which a session reconnect makes it do.
|
|
10
|
-
*/
|
|
11
|
-
import type { PipecatClient } from "@pipecat-ai/client-js";
|
|
12
|
-
import { type AvatarApi } from "../../src/avatar.js";
|
|
13
|
-
export interface UseAvatarOptions {
|
|
14
|
-
/** Which face. Omit for the widget's own default. */
|
|
15
|
-
avatar?: string;
|
|
16
|
-
/** The live `PipecatClient` to dispatch server-messages from, or `null`
|
|
17
|
-
* before connect. `useAvatar` (dis)connects the subscription as this
|
|
18
|
-
* changes; it does not create or own the client. */
|
|
19
|
-
client?: PipecatClient | null;
|
|
20
|
-
}
|
|
21
|
-
/**
|
|
22
|
-
* The mount ref's type, written out rather than named as React's `RefObject`.
|
|
23
|
-
*
|
|
24
|
-
* React 18 and 19 declare that alias with different type arguments — 18's
|
|
25
|
-
* `useRef<T>(null)` yields `RefObject<T>`, 19's yields `RefObject<T | null>` —
|
|
26
|
-
* and because both are the *same alias*, TypeScript compares them by variance
|
|
27
|
-
* and rejects whichever one we didn't pick. An anonymous shape forces a
|
|
28
|
-
* structural comparison instead, which both versions satisfy, and which the
|
|
29
|
-
* `ref` prop accepts on both. This is the only place the 18-vs-19 split shows
|
|
30
|
-
* up in the binding; keep it that way.
|
|
31
|
-
*/
|
|
32
|
-
export type AvatarMountRef = {
|
|
33
|
-
current: HTMLDivElement | null;
|
|
34
|
-
};
|
|
35
|
-
export interface UseAvatarHandle {
|
|
36
|
-
/** Attach to the mount element: `<div ref={containerRef} />`. */
|
|
37
|
-
containerRef: AvatarMountRef;
|
|
38
|
-
/** The live widget instance once mounted, else `null`. */
|
|
39
|
-
avatar: AvatarApi | null;
|
|
40
|
-
}
|
|
41
|
-
export declare function useAvatar(options?: UseAvatarOptions): UseAvatarHandle;
|
|
42
|
-
//# sourceMappingURL=useAvatar.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"useAvatar.d.ts","sourceRoot":"","sources":["../src/useAvatar.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAGH,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,uBAAuB,CAAC;AAC3D,OAAO,EAAgB,KAAK,SAAS,EAAE,MAAM,qBAAqB,CAAC;AAGnE,MAAM,WAAW,gBAAgB;IAC/B,qDAAqD;IACrD,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB;;wDAEoD;IACpD,MAAM,CAAC,EAAE,aAAa,GAAG,IAAI,CAAC;CAC/B;AAED;;;;;;;;;;GAUG;AACH,MAAM,MAAM,cAAc,GAAG;IAAE,OAAO,EAAE,cAAc,GAAG,IAAI,CAAA;CAAE,CAAC;AAEhE,MAAM,WAAW,eAAe;IAC9B,iEAAiE;IACjE,YAAY,EAAE,cAAc,CAAC;IAC7B,0DAA0D;IAC1D,MAAM,EAAE,SAAS,GAAG,IAAI,CAAC;CAC1B;AAED,wBAAgB,SAAS,CAAC,OAAO,GAAE,gBAAqB,GAAG,eAAe,CAuCzE"}
|
package/client/dist/useAvatar.js
DELETED
|
@@ -1,51 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* useAvatar — mount the widget, wire it to a live session, dispatch its
|
|
3
|
-
* server-messages, and clean up.
|
|
4
|
-
*
|
|
5
|
-
* Internal: `<Avatar>` is the only thing the package exports. It is a separate
|
|
6
|
-
* module anyway because the two lifecycles genuinely differ — mounting the
|
|
7
|
-
* widget happens once (an avatar swap remounts by design; see the note in the
|
|
8
|
-
* effect), while attaching to the pipecat client re-runs whenever the client
|
|
9
|
-
* identity changes, which a session reconnect makes it do.
|
|
10
|
-
*/
|
|
11
|
-
import { useEffect, useRef, useState } from "react";
|
|
12
|
-
import { createAvatar } from "../../src/avatar.js";
|
|
13
|
-
import { AvatarClient } from "./AvatarClient.js";
|
|
14
|
-
export function useAvatar(options = {}) {
|
|
15
|
-
const containerRef = useRef(null);
|
|
16
|
-
const [avatar, setAvatar] = useState(null);
|
|
17
|
-
const avatarClientRef = useRef(null);
|
|
18
|
-
// Latest-options ref, so the mount effect (which runs once) still reads the
|
|
19
|
-
// live props without re-subscribing.
|
|
20
|
-
const optionsRef = useRef(options);
|
|
21
|
-
optionsRef.current = options;
|
|
22
|
-
useEffect(() => {
|
|
23
|
-
const mount = containerRef.current;
|
|
24
|
-
if (!mount)
|
|
25
|
-
return;
|
|
26
|
-
const instance = createAvatar({ mount, avatar: optionsRef.current.avatar });
|
|
27
|
-
avatarClientRef.current = new AvatarClient(instance);
|
|
28
|
-
setAvatar(instance);
|
|
29
|
-
return () => {
|
|
30
|
-
instance.destroy();
|
|
31
|
-
avatarClientRef.current = null;
|
|
32
|
-
setAvatar(null);
|
|
33
|
-
};
|
|
34
|
-
// Mount once. `avatar` is read at mount time only — the widget has no
|
|
35
|
-
// hot-swap-avatar API (`createFace` runs once per mount), so changing it
|
|
36
|
-
// re-renders nothing here by design; a caller that needs a different face
|
|
37
|
-
// remounts with a `key` prop (see the component's doc).
|
|
38
|
-
// eslint-disable-next-line react-hooks/exhaustive-deps
|
|
39
|
-
}, []);
|
|
40
|
-
useEffect(() => {
|
|
41
|
-
const wrapper = avatarClientRef.current;
|
|
42
|
-
const pipecatClient = options.client;
|
|
43
|
-
if (!wrapper || !pipecatClient)
|
|
44
|
-
return;
|
|
45
|
-
return wrapper.attach(pipecatClient);
|
|
46
|
-
// Re-subscribe when the widget mounts or the session's client changes.
|
|
47
|
-
// eslint-disable-next-line react-hooks/exhaustive-deps
|
|
48
|
-
}, [avatar, options.client]);
|
|
49
|
-
return { containerRef, avatar };
|
|
50
|
-
}
|
|
51
|
-
//# sourceMappingURL=useAvatar.js.map
|