revert faster-whisper to v1.0.3

.dockerignore

@@ -0,0 +1 @@
+/venv

Dockerfile

@@ -1,7 +1,7 @@
 FROM docker.io/nvidia/cuda:12.0.0-cudnn8-runtime-ubuntu22.04
 
 RUN apt-get update && \
-    apt-get install -y ffmpeg python3 python3-pip && \
+    apt-get install -y ffmpeg python3 python3-pip git && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/*
 

requirements.txt

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

requirements_version.txt

@@ -1,48 +1,49 @@
-annotated-types==0.7.0
+annotated-types==0.7.0
-anyio==4.4.0
+anyio==4.4.0
-av==12.2.0
+av==12.3.0
-certifi==2024.7.4
+certifi==2024.8.30
-cffi==1.16.0
+cffi==1.17.1
-charset-normalizer==3.3.2
+charset-normalizer==3.3.2
-click==8.1.7
+click==8.1.7
-coloredlogs==15.0.1
+coloredlogs==15.0.1
-ctranslate2==4.3.1
+ctranslate2==4.4.0
-fastapi==0.111.0
+exceptiongroup==1.2.2
-faster-whisper==1.0.3
+fastapi==0.114.1
-filelock==3.15.4
+faster-whisper @ git+https://github.com/heimoshuiyu/faster-whisper@28a4d11a736d8cdeb4655ee5d7e4b4e7ae5ec8e0
-flatbuffers==24.3.25
+filelock==3.16.0
-fsspec==2024.6.1
+flatbuffers==24.3.25
-h11==0.14.0
+fsspec==2024.9.0
-httptools==0.6.1
+h11==0.14.0
-huggingface-hub==0.23.4
+httptools==0.6.1
-humanfriendly==10.0
+huggingface-hub==0.24.6
-idna==3.7
+humanfriendly==10.0
-mpmath==1.3.0
+idna==3.8
-numpy==1.26.4
+mpmath==1.3.0
-onnxruntime==1.18.1
+numpy==2.1.1
-OpenCC==1.1.7
+onnxruntime==1.19.2
-packaging==24.1
+OpenCC==1.1.9
-prometheus-client==0.18.0
+packaging==24.1
-prometheus-fastapi-instrumentator==7.0.0
+prometheus-fastapi-instrumentator==7.0.0
-protobuf==5.27.2
+prometheus_client==0.20.0
-pycparser==2.22
+protobuf==5.28.0
-pydantic==2.8.2
+pycparser==2.22
-pydantic_core==2.20.1
+pydantic==2.9.1
-pydub==0.25.1
+pydantic_core==2.23.3
-python-dotenv==1.0.1
+pydub==0.25.1
-python-multipart==0.0.9
+python-dotenv==1.0.1
-PyYAML==6.0.1
+python-multipart==0.0.9
-requests==2.32.3
+PyYAML==6.0.2
-sniffio==1.3.1
+requests==2.32.3
-sounddevice==0.4.7
+sniffio==1.3.1
-starlette==0.37.2
+sounddevice==0.5.0
-sympy==1.12.1
+starlette==0.38.5
-tokenizers==0.19.1
+sympy==1.13.2
-tqdm==4.66.4
+tokenizers==0.20.0
-typing_extensions==4.12.2
+tqdm==4.66.5
-urllib3==2.2.2
+typing_extensions==4.12.2
-uvicorn==0.30.1
+urllib3==2.2.2
-uvloop==0.19.0
+uvicorn==0.30.6
-watchfiles==0.22.0
+uvloop==0.20.0
-websockets==12.0
+watchfiles==0.24.0
-whisper-ctranslate2==0.4.5
+websockets==13.0.1
+whisper-ctranslate2==0.4.5

start-podman.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+
+podman run -d --name whisper-fastapi \
+    --restart unless-stopped \
+    --name whisper-fastapi \
+    -v ~/.cache/huggingface:/root/.cache/huggingface \
+    --device nvidia.com/gpu=all --security-opt=label=disable \
+    --gpus all \
+    -p 5000:5000 \
+    docker.io/heimoshuiyu/whisper-fastapi:latest \
+    --model large-v2

whisper_fastapi.py

@@ -78,7 +78,7 @@ def tsv_writer(generator: Generator[dict[str, Any], Any, None]):
     for i, segment in enumerate(generator):
         start_time = str(round(1000 * segment["start"]))
         end_time = str(round(1000 * segment["end"]))
-        text = segment["text"]
+        text = segment["text"].strip()
         yield f"{start_time}\t{end_time}\t{text}\n"
 
 
@@ -90,7 +90,7 @@ def srt_writer(generator: Generator[dict[str, Any], Any, None]):
         end_time = format_timestamp(
             segment["end"], decimal_marker=",", always_include_hours=True
         )
-        text = segment["text"]
+        text = segment["text"].strip()
         yield f"{i}\n{start_time} --> {end_time}\n{text}\n\n"
 
 
@@ -99,7 +99,7 @@ def vtt_writer(generator: Generator[dict[str, Any], Any, None]):
     for i, segment in enumerate(generator):
         start_time = format_timestamp(segment["start"])
         end_time = format_timestamp(segment["end"])
-        text = segment["text"]
+        text = segment["text"].strip()
         yield f"{start_time} --> {end_time}\n{text}\n\n"
 
 
@@ -159,6 +159,8 @@ def stream_builder(
 
 @app.websocket("/k6nele/status")
 @app.websocket("/konele/status")
+@app.websocket("/v1/k6nele/status")
+@app.websocket("/v1/konele/status")
 async def konele_status(
     websocket: WebSocket,
 ):
@@ -169,6 +171,8 @@ async def konele_status(
 
 @app.websocket("/k6nele/ws")
 @app.websocket("/konele/ws")
+@app.websocket("/v1/k6nele/ws")
+@app.websocket("/v1/konele/ws")
 async def konele_ws(
     websocket: WebSocket,
     task: Literal["transcribe", "translate"] = "transcribe",
@@ -233,6 +237,8 @@ async def konele_ws(
 
 @app.post("/k6nele/post")
 @app.post("/konele/post")
+@app.post("/v1/k6nele/post")
+@app.post("/v1/konele/post")
 async def translateapi(
     request: Request,
     task: Literal["transcribe", "translate"] = "transcribe",