Commit @0e0a12e470c70bc2b475fe0c10f11846f1dc74ab - yjyoon/whisper_streaming

Quentin Fuxa 2024-12-29

Batch unprocessed audio to reduce Whisper streaming calls

@0e0a12e470c70bc2b475fe0c10f11846f1dc74ab

29ed46d

0e0a12e

whisper_fastapi_online_server.py

--- whisper_fastapi_online_server.py

+++ whisper_fastapi_online_server.py


 import asyncio
 import numpy as np
 import ffmpeg
+from time import time
 
 from fastapi import FastAPI, WebSocket, WebSocketDisconnect
 from fastapi.responses import HTMLResponse

         nonlocal pcm_buffer
         loop = asyncio.get_event_loop()
         full_transcription = ""
+        beg = time()
         while True:
             try:
-                chunk = await loop.run_in_executor(None, ffmpeg_process.stdout.read, 4096)
-                if not chunk:  # FFmpeg might have closed
-                    print("FFmpeg stdout closed.")
-                    break
+                elapsed_time = int(time() - beg)
+                beg = time()
+                chunk = await loop.run_in_executor(None, ffmpeg_process.stdout.read, 32000*elapsed_time)
+                if not chunk: # The first chunk will be almost empty, FFmpeg is still starting up
+                    chunk = await loop.run_in_executor(None, ffmpeg_process.stdout.read, 4096)
+                    if not chunk: # FFmpeg might have closed
+                        print("FFmpeg stdout closed.")
+                        break
 
                 pcm_buffer.extend(chunk)
 
-                while len(pcm_buffer) >= BYTES_PER_SEC:
-                    three_sec_chunk = pcm_buffer[:BYTES_PER_SEC]
-                    del pcm_buffer[:BYTES_PER_SEC]
-
+                if len(pcm_buffer) >= BYTES_PER_SEC:
                     # Convert int16 -> float32
-                    pcm_array = np.frombuffer(three_sec_chunk, dtype=np.int16).astype(np.float32) / 32768.0
-
+                    pcm_array = np.frombuffer(pcm_buffer, dtype=np.int16).astype(np.float32) / 32768.0
+                    pcm_buffer = bytearray()
                     online.insert_audio_chunk(pcm_array)
                     transcription = online.process_iter()[2]
                     if args.vac:

Add a comment

Open 0
Closed 0

List

...	...	@@ -3,6 +3,7 @@
3	3	import asyncio
4	4	import numpy as np
5	5	import ffmpeg
	6	+from time import time
6	7
7	8	from fastapi import FastAPI, WebSocket, WebSocketDisconnect
8	9	from fastapi.responses import HTMLResponse
...	...	@@ -69,22 +70,24 @@
69	70	nonlocal pcm_buffer
70	71	loop = asyncio.get_event_loop()
71	72	full_transcription = ""
	73	+ beg = time()
72	74	while True:
73	75	try:
74		- chunk = await loop.run_in_executor(None, ffmpeg_process.stdout.read, 4096)
75		- if not chunk: # FFmpeg might have closed
76		- print("FFmpeg stdout closed.")
77		- break
	76	+ elapsed_time = int(time() - beg)
	77	+ beg = time()
	78	+ chunk = await loop.run_in_executor(None, ffmpeg_process.stdout.read, 32000*elapsed_time)
	79	+ if not chunk: # The first chunk will be almost empty, FFmpeg is still starting up
	80	+ chunk = await loop.run_in_executor(None, ffmpeg_process.stdout.read, 4096)
	81	+ if not chunk: # FFmpeg might have closed
	82	+ print("FFmpeg stdout closed.")
	83	+ break
78	84
79	85	pcm_buffer.extend(chunk)
80	86
81		- while len(pcm_buffer) >= BYTES_PER_SEC:
82		- three_sec_chunk = pcm_buffer[:BYTES_PER_SEC]
83		- del pcm_buffer[:BYTES_PER_SEC]
84		-
	87	+ if len(pcm_buffer) >= BYTES_PER_SEC:
85	88	# Convert int16 -> float32
86		- pcm_array = np.frombuffer(three_sec_chunk, dtype=np.int16).astype(np.float32) / 32768.0
87		-
	89	+ pcm_array = np.frombuffer(pcm_buffer, dtype=np.int16).astype(np.float32) / 32768.0
	90	+ pcm_buffer = bytearray()
88	91	online.insert_audio_chunk(pcm_array)
89	92	transcription = online.process_iter()[2]
90	93	if args.vac:

Delete comment