code (#48)

freddyaboulton · Jan 10, 2025 · 4d16634 · 4d16634
1 parent b64e019
commit 4d16634
Show file tree

Hide file tree

Showing 12 changed files with 431 additions and 103 deletions.
diff --git a/backend/gradio_webrtc/__init__.py b/backend/gradio_webrtc/__init__.py
@@ -16,10 +16,21 @@
     audio_to_file,
     audio_to_float32,
 )
-from .webrtc import AsyncStreamHandler, StreamHandler, WebRTC
+from .webrtc import (
+    AsyncAudioVideoStreamHandler,
+    AsyncStreamHandler,
+    AudioVideoStreamHandler,
+    StreamHandler,
+    WebRTC,
+    VideoEmitType,
+    AudioEmitType,
+)
 
 __all__ = [
     "AsyncStreamHandler",
+    "AudioVideoStreamHandler",
+    "AudioEmitType",
+    "AsyncAudioVideoStreamHandler",
     "AlgoOptions",
     "AdditionalOutputs",
     "aggregate_bytes_to_16bit",
@@ -36,6 +47,7 @@
     "stt",
     "stt_for_chunks",
     "StreamHandler",
+    "VideoEmitType",
     "WebRTC",
     "WebRTCError",
     "Warning",

diff --git a/backend/gradio_webrtc/reply_on_pause.py b/backend/gradio_webrtc/reply_on_pause.py
@@ -9,7 +9,6 @@
 import numpy as np
 
 from gradio_webrtc.pause_detection import SileroVADModel, SileroVadOptions
-from gradio_webrtc.utils import AdditionalOutputs
 from gradio_webrtc.webrtc import EmitType, StreamHandler
 
 logger = getLogger(__name__)

diff --git a/backend/gradio_webrtc/utils.py b/backend/gradio_webrtc/utils.py
@@ -147,16 +147,18 @@ async def player_worker_decode(
 
             logger.debug(
                 "received array with shape %s sample rate %s layout %s",
-                audio_array.shape,
+                audio_array.shape,  # type: ignore
                 sample_rate,
-                layout,
+                layout,  # type: ignore
             )
-            format = "s16" if audio_array.dtype == "int16" else "fltp"
+            format = "s16" if audio_array.dtype == "int16" else "fltp"  # type: ignore
 
             # Convert to audio frame and resample
             # This runs in the same timeout context
             frame = av.AudioFrame.from_ndarray(  # type: ignore
-                audio_array, format=format, layout=layout
+                audio_array,  # type: ignore
+                format=format,
+                layout=layout,  # type: ignore
             )
             frame.sample_rate = sample_rate