request microphone permission, fix librosa parameters, add libsndfile1 system dependency, fix list itterator, handle collect_frames asyncio task cancelations

Author: eliteprox

docker/Dockerfile.base

@@ -6,6 +6,7 @@ RUN export DEBIAN_FRONTEND=noninteractive && apt-get update && apt-get install -
     wget \
     nano \
     socat \
+    libsndfile1 \
     build-essential llvm tk-dev \
     && rm -rf /var/lib/apt/lists/*
 

nodes/audio_utils/pitch_shift.py

@@ -27,6 +27,6 @@ def IS_CHANGED(cls):
 
     def execute(self, audio, sample_rate, pitch_shift):
         audio_float = audio.astype(np.float32) / 32768.0
-        shifted_audio = librosa.effects.pitch_shift(audio_float, sample_rate, n_steps=pitch_shift)
+        shifted_audio = librosa.effects.pitch_shift(y=audio_float, sr=sample_rate, n_steps=pitch_shift)
         shifted_int16 = np.clip(shifted_audio * 32768.0, -32768, 32767).astype(np.int16)
         return shifted_int16, sample_rate

server/app.py

@@ -22,6 +22,7 @@
 logger = logging.getLogger(__name__)
 logging.getLogger('aiortc.rtcrtpsender').setLevel(logging.WARNING)
 logging.getLogger('aiortc.rtcrtpreceiver').setLevel(logging.WARNING)
+logging.getLogger('numba').setLevel(logging.WARNING)
 
 
 MAX_BITRATE = 2000000
@@ -37,9 +38,12 @@ def __init__(self, track: MediaStreamTrack, pipeline):
         asyncio.create_task(self.collect_frames())
 
     async def collect_frames(self):
-        while True:
-            frame = await self.track.recv()
-            await self.pipeline.put_video_frame(frame)
+        try:
+            while True:
+                frame = await self.track.recv()
+                await self.pipeline.put_video_frame(frame)
+        except asyncio.CancelledError:
+            logger.info("Audio frame collection cancelled")
 
     async def recv(self):
         return await self.pipeline.get_processed_video_frame()
@@ -54,9 +58,12 @@ def __init__(self, track: MediaStreamTrack, pipeline):
         asyncio.create_task(self.collect_frames())
 
     async def collect_frames(self):
-        while True:
-            frame = await self.track.recv()
-            await self.pipeline.put_audio_frame(frame)
+        try:
+            while True:
+                frame = await self.track.recv()
+                await self.pipeline.put_audio_frame(frame)
+        except asyncio.CancelledError:
+            logger.info("Audio frame collection cancelled")
 
     async def recv(self):
         return await self.pipeline.get_processed_audio_frame()

src/comfystream/client.py

@@ -22,7 +22,7 @@ def __init__(self, max_workers: int = 1, **kwargs):
 
     async def set_prompts(self, prompts: List[PromptDictInput]):
         self.current_prompts = [convert_prompt(prompt) for prompt in prompts]
-        for idx in range(self.current_prompts):
+        for idx in range(len(self.current_prompts)):
             task = asyncio.create_task(self.run_prompt(idx))
             self.running_prompts[idx] = task
 

ui/src/components/settings.tsx

@@ -146,7 +146,7 @@ function ConfigForm({ config, onSubmit }: ConfigFormProps) {
 
   const getVideoDevices = useCallback(async () => {
     try {
-      await navigator.mediaDevices.getUserMedia({ video: true });
+      await navigator.mediaDevices.getUserMedia({ video: true, audio: true });
 
       const devices = await navigator.mediaDevices.enumerateDevices();
       const videoDevices = [