Labyricorn/voicebox
1
0
Fork 0
mirror of https://github.com/jamiepine/voicebox.git synced 2026-09-19 23:00:45 -07:00
Code Issues Packages Projects Releases Wiki Activity

feat: add download cancel/clear UI, fix whisper-large and error reporting

Browse Source 
- Add cancel (X) button on downloading and errored model items
- Add collapsible Problems panel (VS Code-style) showing error details
- Add "Clear All" button to reset all stale download/error state
- Add POST /models/download/cancel endpoint to dismiss individual downloads
- Add POST /tasks/clear endpoint to reset all task and progress state
- Include error messages in /tasks/active response for visibility
- Capture SSE error messages client-side for immediate display
- Fix whisper-large using wrong HF repo (openai/whisper-large → openai/whisper-large-v3)
- Fix Whisper HF repo mapping in both PyTorch and MLX backends
- Shorten error toast to point users to Problems panel instead of wall of text
This commit is contained in:
Daddy Raegen
2026-03-06 00:56:14 -05:00
parent 38bf96ff20
commit a362d7de2a
9 changed files with 301 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/backends/mlx_backend.py
+12 -4
View File
@@ -379,9 +379,17 @@ class MLXTTSBackend:
return audio, sample_rate
WHISPER_HF_REPOS = {
"base": "openai/whisper-base",
"small": "openai/whisper-small",
"medium": "openai/whisper-medium",
"large": "openai/whisper-large-v3",
}
class MLXSTTBackend:
"""MLX-based STT backend using mlx-audio Whisper."""
def __init__(self, model_size: str = "base"):
self.model = None
self.model_size = model_size
@@ -402,8 +410,8 @@ class MLXSTTBackend:
"""
try:
from huggingface_hub import constants as hf_constants
model_name = f"openai/whisper-{model_size}"
repo_cache = Path(hf_constants.HF_HUB_CACHE) / ("models--" + model_name.replace("/", "--"))
hf_repo = WHISPER_HF_REPOS.get(model_size, f"openai/whisper-{model_size}")
repo_cache = Path(hf_constants.HF_HUB_CACHE) / ("models--" + hf_repo.replace("/", "--"))
if not repo_cache.exists():
return False
@@ -474,7 +482,7 @@ class MLXSTTBackend:
from mlx_audio.stt import load
# MLX Whisper uses the standard OpenAI models
model_name = f"openai/whisper-{model_size}"
model_name = WHISPER_HF_REPOS.get(model_size, f"openai/whisper-{model_size}")
print(f"Loading MLX Whisper model {model_size}...")
backend/backends/pytorch_backend.py
+17 -9
View File
@@ -369,9 +369,17 @@ class PyTorchTTSBackend:
return audio, sample_rate
WHISPER_HF_REPOS = {
"base": "openai/whisper-base",
"small": "openai/whisper-small",
"medium": "openai/whisper-medium",
"large": "openai/whisper-large-v3",
}
class PyTorchSTTBackend:
"""PyTorch-based STT backend using Whisper."""
def __init__(self, model_size: str = "base"):
self.model = None
self.processor = None
@@ -416,18 +424,18 @@ class PyTorchSTTBackend:
"""
try:
from huggingface_hub import constants as hf_constants
model_name = f"openai/whisper-{model_size}"
repo_cache = Path(hf_constants.HF_HUB_CACHE) / ("models--" + model_name.replace("/", "--"))
hf_repo = WHISPER_HF_REPOS.get(model_size, f"openai/whisper-{model_size}")
repo_cache = Path(hf_constants.HF_HUB_CACHE) / ("models--" + hf_repo.replace("/", "--"))
if not repo_cache.exists():
return False
# Check for .incomplete files - if any exist, download is still in progress
blobs_dir = repo_cache / "blobs"
if blobs_dir.exists() and any(blobs_dir.glob("*.incomplete")):
print(f"[_is_model_cached] Found .incomplete files for whisper-{model_size}, treating as not cached")
return False
# Check that actual model weight files exist in snapshots
snapshots_dir = repo_cache / "snapshots"
if snapshots_dir.exists():
@@ -438,12 +446,12 @@ class PyTorchSTTBackend:
if not has_weights:
print(f"[_is_model_cached] No model weights found for whisper-{model_size}, treating as not cached")
return False
return True
except Exception as e:
print(f"[_is_model_cached] Error checking cache for whisper-{model_size}: {e}")
return False
async def load_model_async(self, model_size: Optional[str] = None):
"""
Lazy load the Whisper model.
@@ -494,7 +502,7 @@ class PyTorchSTTBackend:
# Import transformers
from transformers import WhisperProcessor, WhisperForConditionalGeneration
model_name = f"openai/whisper-{model_size}"
model_name = WHISPER_HF_REPOS.get(model_size, f"openai/whisper-{model_size}")
print(f"[DEBUG] Model name: {model_name}")
print(f"Loading Whisper model {model_size} on {self.device}...")
backend/main.py
+45 -4
View File
@@ -1310,14 +1310,14 @@ async def get_model_status():
whisper_base_id = "openai/whisper-base"
whisper_small_id = "openai/whisper-small"
whisper_medium_id = "openai/whisper-medium"
whisper_large_id = "openai/whisper-large"
whisper_large_id = "openai/whisper-large-v3"
else:
tts_1_7b_id = "Qwen/Qwen3-TTS-12Hz-1.7B-Base"
tts_0_6b_id = "Qwen/Qwen3-TTS-12Hz-0.6B-Base"
whisper_base_id = "openai/whisper-base"
whisper_small_id = "openai/whisper-small"
whisper_medium_id = "openai/whisper-medium"
whisper_large_id = "openai/whisper-large"
whisper_large_id = "openai/whisper-large-v3"
model_configs = [
{
@@ -1586,6 +1586,39 @@ async def trigger_model_download(request: models.ModelDownloadRequest):
return {"message": f"Model {request.model_name} download started"}
@app.post("/models/download/cancel")
async def cancel_model_download(request: models.ModelDownloadRequest):
"""Cancel or dismiss an errored/stale download task."""
task_manager = get_task_manager()
progress_manager = get_progress_manager()
removed = task_manager.cancel_download(request.model_name)
# Also clear progress state so the model doesn't show as downloading
with progress_manager._lock:
if request.model_name in progress_manager._progress:
del progress_manager._progress[request.model_name]
return {"message": f"Download task for {request.model_name} cancelled"}
@app.post("/tasks/clear")
async def clear_all_tasks():
"""Clear all download tasks and progress state. Does not delete downloaded files."""
task_manager = get_task_manager()
progress_manager = get_progress_manager()
task_manager._active_downloads.clear()
task_manager._active_generations.clear()
with progress_manager._lock:
progress_manager._progress.clear()
progress_manager._last_notify_time.clear()
progress_manager._last_notify_progress.clear()
return {"message": "All task state cleared"}
@app.delete("/models/{model_name}")
async def delete_model(model_name: str):
"""Delete a downloaded model from the HuggingFace cache."""
@@ -1621,12 +1654,12 @@ async def delete_model(model_name: str):
"model_type": "whisper",
},
"whisper-large": {
"hf_repo_id": "openai/whisper-large",
"hf_repo_id": "openai/whisper-large-v3",
"model_size": "large",
"model_type": "whisper",
},
}
if model_name not in model_configs:
raise HTTPException(status_code=400, detail=f"Unknown model: {model_name}")
@@ -1710,10 +1743,18 @@ async def get_active_tasks():
progress = progress_map.get(model_name)
if task:
# Prefer task error, fall back to progress manager error
error = task.error
if not error:
with progress_manager._lock:
pm_data = progress_manager._progress.get(model_name)
if pm_data:
error = pm_data.get("error")
active_downloads.append(models.ActiveDownloadTask(
model_name=model_name,
status=task.status,
started_at=task.started_at,
error=error,
))
elif progress:
# Progress exists but no task - create from progress data
backend/models.py
+1
View File
@@ -154,6 +154,7 @@ class ActiveDownloadTask(BaseModel):
model_name: str
status: str
started_at: datetime
error: Optional[str] = None
class ActiveGenerationTask(BaseModel):
backend/utils/tasks.py
+7
View File
@@ -72,6 +72,13 @@ class TaskManager:
"""Get all active generations."""
return list(self._active_generations.values())
def cancel_download(self, model_name: str) -> bool:
"""Cancel/dismiss a download task (removes it from active list)."""
if model_name in self._active_downloads:
del self._active_downloads[model_name]
return True
return False
def is_download_active(self, model_name: str) -> bool:
"""Check if a download is active."""
return model_name in self._active_downloads