Labyricorn/voicebox
1
0
Fork 0
mirror of https://github.com/jamiepine/voicebox.git synced 2026-10-04 09:35:16 -07:00
Code Issues Packages Projects Releases Wiki Activity

fix(backend): count the S3TokenizerV2 repo in the Chatterbox MLX cache check; register the backend with PyInstaller

Browse Source 
Review follow-ups: mlx-audio's Model.from_pretrained fetches the S3
speech tokenizer from mlx-community/S3TokenizerV2 (~470 MB) separately
from the chatterbox checkout, so _is_model_cached now requires both
repos (same shape as the Hume backend's codec check) and the config's
size_mb reflects the real footprint. backend.backends.chatterbox_mlx_backend
is a function-level import that PyInstaller's graph will not see, so it
is added to the Apple Silicon hidden-import list in build_binary.py and
voicebox-server.spec next to mlx_backend.
This commit is contained in:
jamiepine
2026-10-04 00:25:58 +00:00
committed by capy-ai-staging[bot]
parent c5cf7436f7
commit ad6ec3c6ef
4 changed files with 13 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/backends/__init__.py
+2 -1
View File
@@ -297,7 +297,8 @@ def _get_non_qwen_tts_configs() -> list[ModelConfig]:
on_mlx = get_backend_type() == "mlx"
if on_mlx:
chatterbox_repo = "mlx-community/chatterbox-multilingual-v3"
chatterbox_size_mb = 2600
# 2.5 GB of weights plus the separately fetched S3TokenizerV2 (~470 MB)
chatterbox_size_mb = 3000
else:
chatterbox_repo = "ResembleAI/chatterbox"
chatterbox_size_mb = 3200
backend/backends/chatterbox_mlx_backend.py
+8 -2
View File
@@ -30,9 +30,13 @@ from .mlx_backend import _run_on_mlx_thread
logger = logging.getLogger(__name__)
CHATTERBOX_MLX_HF_REPO = "mlx-community/chatterbox-multilingual-v3"
# mlx-audio's Model.from_pretrained fetches the S3 speech tokenizer from this
# second repo (~470 MB), so the engine is only "downloaded" once both are cached.
S3_TOKENIZER_HF_REPO = "mlx-community/S3TokenizerV2"
# Files that must be present for the MLX multilingual model
_MLX_WEIGHT_FILES = ["model.safetensors", "config.json", "tokenizer.json"]
_S3_TOKENIZER_FILES = ["model.safetensors", "config.json"]
class ChatterboxMLXTTSBackend:
@@ -51,7 +55,9 @@ class ChatterboxMLXTTSBackend:
return CHATTERBOX_MLX_HF_REPO
def _is_model_cached(self, model_size: str = "default") -> bool:
return is_model_cached(CHATTERBOX_MLX_HF_REPO, required_files=_MLX_WEIGHT_FILES)
model_cached = is_model_cached(CHATTERBOX_MLX_HF_REPO, required_files=_MLX_WEIGHT_FILES)
tokenizer_cached = is_model_cached(S3_TOKENIZER_HF_REPO, required_files=_S3_TOKENIZER_FILES)
return model_cached and tokenizer_cached
async def load_model(self, model_size: str = "default") -> None:
"""Load the Chatterbox multilingual MLX model."""
@@ -75,7 +81,7 @@ class ChatterboxMLXTTSBackend:
logger.info("Loading Chatterbox Multilingual TTS on MLX (Metal)...")
ckpt_dir = snapshot_download(CHATTERBOX_MLX_HF_REPO)
self.model = Model.from_pretrained(ckpt_dir)
self.model = Model.from_pretrained(ckpt_dir, s3_tokenizer_repo=S3_TOKENIZER_HF_REPO)
logger.info("Chatterbox Multilingual TTS (MLX) loaded successfully")