heimoshuiyu/whisper-fastapi
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Revert "disable zh-hans to zh-cn convertion"

Browse Source 
This reverts commit 6decabefae.
This commit is contained in:
黑墨水鱼
2024-08-04 16:39:20 +08:00
parent d86ed9be69
commit 204ccb8f3d
1 changed files with 8 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
whisper_fastapi.py
View File

@@ -22,6 +22,7 @@ from fastapi.middleware.cors import CORSMiddleware
from src.whisper_ctranslate2.whisper_ctranslate2 import Transcribe from src.whisper_ctranslate2.whisper_ctranslate2 import Transcribe
from src.whisper_ctranslate2.writers import format_timestamp from src.whisper_ctranslate2.writers import format_timestamp
from faster_whisper.transcribe import Segment, TranscriptionInfo from faster_whisper.transcribe import Segment, TranscriptionInfo
import opencc
from prometheus_fastapi_instrumentator import Instrumentator from prometheus_fastapi_instrumentator import Instrumentator
parser = argparse.ArgumentParser() parser = argparse.ArgumentParser()
@@ -36,6 +37,7 @@ args = parser.parse_args()
app = FastAPI() app = FastAPI()
# Instrument your app with default metrics and expose the metrics # Instrument your app with default metrics and expose the metrics
Instrumentator().instrument(app).expose(app, endpoint="/konele/metrics") Instrumentator().instrument(app).expose(app, endpoint="/konele/metrics")
ccc = opencc.OpenCC("t2s.json")
print("Loading model...") print("Loading model...")
transcriber = Transcribe( transcriber = Transcribe(
@@ -142,6 +144,7 @@ def stream_builder(
data = segment._asdict() data = segment._asdict()
if data.get('words') is not None: if data.get('words') is not None:
data["words"] = [i._asdict() for i in data["words"]] data["words"] = [i._asdict() for i in data["words"]]
data["text"] = ccc.convert(data["text"])
yield data yield data
info_dict = info._asdict() info_dict = info._asdict()
@@ -209,14 +212,14 @@ async def konele_ws(
file_obj.seek(0) file_obj.seek(0)
generator, info = stream_builder( generator = stream_builder(
audio=file_obj, audio=file_obj,
task=task, task=task,
vad_filter=vad_filter, vad_filter=vad_filter,
language=None if lang == "und" else lang, language=None if lang == "und" else lang,
initial_prompt=initial_prompt, initial_prompt=initial_prompt,
) )
result = build_json_result(generator, info) result = build_json_result(generator)
text = result.get("text", "") text = result.get("text", "")
print("result", text) print("result", text)
@@ -273,14 +276,14 @@ async def translateapi(
file_obj.seek(0) file_obj.seek(0)
generator, info = stream_builder( generator = stream_builder(
audio=file_obj, audio=file_obj,
task=task, task=task,
vad_filter=vad_filter, vad_filter=vad_filter,
language=None if lang == "und" else lang, language=None if lang == "und" else lang,
initial_prompt=initial_prompt, initial_prompt=initial_prompt,
) )
result = build_json_result(generator, info) result = build_json_result(generator)
text = result.get("text", "") text = result.get("text", "")
print("result", text) print("result", text)
@@ -323,7 +326,7 @@ async def transcription(
media_type="text/event-stream", media_type="text/event-stream",
) )
elif response_format == "json": elif response_format == "json":
return build_json_result(generator, info) return build_json_result(generator)
elif response_format == "text": elif response_format == "text":
return StreamingResponse(text_writer(generator), media_type="text/plain") return StreamingResponse(text_writer(generator), media_type="text/plain")
elif response_format == "tsv": elif response_format == "tsv":