update faster-whisper

requirements.txt

@@ -4,5 +4,5 @@ uvicorn[standard]
 whisper_ctranslate2
 opencc
 prometheus-fastapi-instrumentator
-git+https://github.com/SYSTRAN/faster-whisper@be9fb36ed356b9e299b125de6ee91862e0ac9038
+git+https://github.com/heimoshuiyu/faster-whisper@a759f5f48f5ef5b79461a6461966eafe9df088a9
 pydub

requirements_version.txt

@@ -9,7 +9,7 @@ coloredlogs==15.0.1
 ctranslate2==4.5.0
 exceptiongroup==1.2.2
 fastapi==0.115.5
-faster-whisper @ git+https://github.com/SYSTRAN/faster-whisper@be9fb36ed356b9e299b125de6ee91862e0ac9038
+faster-whisper @ git+https://github.com/heimoshuiyu/faster-whisper@a759f5f48f5ef5b79461a6461966eafe9df088a9
 filelock==3.16.1
 flatbuffers==24.3.25
 fsspec==2024.10.0
@@ -20,15 +20,15 @@ humanfriendly==10.0
 idna==3.10
 mpmath==1.3.0
 numpy==2.1.3
-onnxruntime==1.20.0
+onnxruntime==1.20.1
 OpenCC==1.1.9
 packaging==24.2
 prometheus-fastapi-instrumentator==7.0.0
 prometheus_client==0.21.0
 protobuf==5.28.3
 pycparser==2.22
-pydantic==2.9.2
-pydantic_core==2.23.4
+pydantic==2.10.1
+pydantic_core==2.27.1
 pydub==0.25.1
 python-dotenv==1.0.1
 python-multipart==0.0.17
@@ -36,14 +36,14 @@ PyYAML==6.0.2
 requests==2.32.3
 sniffio==1.3.1
 sounddevice==0.5.1
-starlette==0.41.2
+starlette==0.41.3
 sympy==1.13.3
 tokenizers==0.20.3
 tqdm==4.67.0
 typing_extensions==4.12.2
 urllib3==2.2.3
-uvicorn==0.32.0
+uvicorn==0.32.1
 uvloop==0.21.0
 watchfiles==0.24.0
 websockets==14.1
-whisper-ctranslate2==0.4.7
+whisper-ctranslate2==0.4.8

whisper_fastapi.py

@@ -1,6 +1,6 @@
+import sys
 import dataclasses
 import faster_whisper
-import tqdm
 import json
 from fastapi.responses import StreamingResponse
 import wave
@@ -21,12 +21,16 @@ from fastapi import (
     WebSocket,
 )
 from fastapi.middleware.cors import CORSMiddleware
-from src.whisper_ctranslate2.whisper_ctranslate2 import Transcribe
 from src.whisper_ctranslate2.writers import format_timestamp
 from faster_whisper.transcribe import Segment, TranscriptionInfo
 import opencc
 from prometheus_fastapi_instrumentator import Instrumentator
 
+# redirect print to stderr
+_print = print
+def print(*args, **kwargs):
+    _print(*args, file=sys.stderr, **kwargs)
+
 parser = argparse.ArgumentParser()
 parser.add_argument("--host", default="0.0.0.0", type=str)
 parser.add_argument("--port", default=5000, type=int)
@@ -95,7 +99,7 @@ def srt_writer(generator: Generator[Segment, Any, None]):
 
 def vtt_writer(generator: Generator[Segment, Any, None]):
     yield "WEBVTT\n\n"
-    for i, segment in enumerate(generator):
+    for _, segment in enumerate(generator):
         start_time = format_timestamp(segment.start)
         end_time = format_timestamp(segment.end)
         text = segment.text.strip()
@@ -107,15 +111,16 @@ class JsonResult(TranscriptionInfo):
     segments: list[Segment]
     text: str
 
+
 def build_json_result(
     generator: Iterable[Segment],
-    info:  TranscriptionInfo,
+    info: TranscriptionInfo,
 ) -> JsonResult:
     segments = [i for i in generator]
     return JsonResult(
         text="\n".join(i.text for i in segments),
         segments=segments,
-        **dataclasses.asdict(info)
+        **dataclasses.asdict(info),
     )
 
 
@@ -140,14 +145,12 @@ def stream_builder(
         "Detected language '%s' with probability %f"
         % (info.language, info.language_probability)
     )
+
     def wrap():
-        last_pos = 0
-        with tqdm.tqdm(total=info.duration, unit="seconds", disable=True) as pbar:
-            for segment in segments:
-                start, end, text = segment.start, segment.end, segment.text
-                pbar.update(end - last_pos)
-                last_pos = end
-                yield segment
+        for segment in segments:
+            if info.language == "zh":
+                segment.text = ccc.convert(segment.text)
+            yield segment
 
     return wrap(), info
 
@@ -303,9 +306,9 @@ async def transcription(
     """
 
     if not task:
-        if request.url.path == '/v1/audio/transcriptions':
+        if request.url.path == "/v1/audio/transcriptions":
             task = "transcribe"
-        elif request.url.path == '/v1/audio/translations':
+        elif request.url.path == "/v1/audio/translations":
             task = "translate"
         else:
             raise HTTPException(400, "task parameter is required")