0x2620/pandora_transcribe
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

select model

Browse source
This commit is contained in:
j 2025-01-24 11:50:49 +05:30
parent 4853d9bfb7
commit 7f11a62243
1 changed files with 6 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
run_whisper.py
View file

@ -6,8 +6,9 @@ import sys
import whisper_timestamped as whisper import whisper_timestamped as whisper
def transcribe(path, language=None, translate=None): def transcribe(path, language=None, translate=None, model="small"):
model = whisper.load_model("large-v3-turbo") #model = whisper.load_model("large-v3-turbo")
model = whisper.load_model(model)
audio = whisper.load_audio(path) audio = whisper.load_audio(path)
if translate: if translate:
task = "translate" task = "translate"
@ -33,10 +34,11 @@ if __name__ == "__main__":
usage = "usage: %(prog)s [options] path output" usage = "usage: %(prog)s [options] path output"
parser = ArgumentParser(usage=usage, prog="run_whisper") parser = ArgumentParser(usage=usage, prog="run_whisper")
parser.add_argument("-l", "--language", dest="language", default=None) parser.add_argument("-l", "--language", dest="language", default=None)
parser.add_argument("-t", "--translate", dest="translate", default=None) parser.add_argument("-m", "--model", dest="model", default="small")
parser.add_argument("-t", "--translate", dest="translate", action='store_true', default=False)
parser.add_argument("path", metavar="path", type=str, help="media input") parser.add_argument("path", metavar="path", type=str, help="media input")
parser.add_argument("output", metavar="output", type=str, help="json output") parser.add_argument("output", metavar="output", type=str, help="json output")
opts = parser.parse_args() opts = parser.parse_args()
result = transcribe(opts.path, opts.language, opts.translate) result = transcribe(opts.path, opts.language, opts.translate, opts.model)
with open(opts.output, "w") as fd: with open(opts.output, "w") as fd:
json.dump(result, fd, indent=2, ensure_ascii=False) json.dump(result, fd, indent=2, ensure_ascii=False)