avatar_lipsync 0.3.0
avatar_lipsync: ^0.3.0 copied to clipboard
Real-time lip sync for Flutter. Turns streamed speech into mouth shapes and animates a drawn face or a single photo (web, via MediaPipe).
0.3.0 #
RigOverlayonPhotoLipSyncAvatar: draws the triangle mesh, the 478 landmarks and a heat map, either of live motion (with the head's own motion taken out) or of a region's weights (RigRegion: jaw, lips, corners, eyelids, brows, irises, head).FaceRig.weights(region)andFaceRig.landmarkCount.- Example:
MicSpeechSource, aLipSyncSourceover the microphone. - Live demo at https://lipsync.azanello.com (in
demo/, not part of the package).
0.2.0 #
pcm16ToSamples: converts little-endian 16-bit PCM, as OpenAI Realtime and Gemini Live stream it, to samples for aSpeechChunk.- README: recipe for realtime voice models (OpenAI Realtime, Gemini Live).
- Example:
WebAudioSpeechSource.playPcm16, a short fade onstop(), a scheduling lead after gaps, and a latency fix for Safari, which has nooutputLatency.
0.1.1 #
- Example: uses
package:material_ui, sincepackage:flutter/material.dartis deprecated on the current stable Flutter. - Releases are published from GitHub Actions (version tags, OIDC).
0.1.0 #
First release.
LipSyncSource: the interface your audio player implements (scheduled chunks, flushes, audible time).SpeechTimelineandLipSyncAnalyzer: speech analysis ahead of playback and a JALI-style mouth model (jaw, lip gap, width, rounding, press, viseme).LipSyncAvatar: a drawn face for every platform.PhotoLipSyncAvatar: a rigged photo on the web (MediaPipe Face Landmarker loaded at runtime), with blinks, head motion, brows and gaze fromAvatarMotion.