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

Compare commits

base: 0x2620:98c680e89cbe46029ef15f40aaffd252649947ba
Branches Tags
0x2620:main
..
compare: 0x2620:5e53ad8556daf2ad6345ae3b960eaf1661cdb11d
Branches Tags
0x2620:main

No commits in common. "98c680e89cbe46029ef15f40aaffd252649947ba" and "5e53ad8556daf2ad6345ae3b960eaf1661cdb11d" have entirely different histories.
98c680e89c ... 5e53ad8556

3 changed files with 8 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

4
README.md
View file

@ -14,8 +14,8 @@ add "transcribe" to LOCAL_APPS in local_setttings.py
apt install portaudio19-dev apt install portaudio19-dev
python3 -m venv venv /opt/whisper-timestamped/ python3 -m venv venv /opt/whisper-timestamped/
/opt/whisper-timestamped/bin/pip install \ /opt/whisper-timestamped/bin/pip install \
torch==2.3.1+cpu \ torch==1.13.1+cpu \
torchaudio==2.3.1+cpu \ torchaudio==0.13.1+cpu \
-f https://download.pytorch.org/whl/torch_stable.html -f https://download.pytorch.org/whl/torch_stable.html
/opt/whisper-timestamped/bin/pip install whisper-timestamped transformers auditok /opt/whisper-timestamped/bin/pip install whisper-timestamped transformers auditok

10
run_whisper.py
View file

@ -6,9 +6,8 @@ import sys
import whisper_timestamped as whisper import whisper_timestamped as whisper
def transcribe(path, language=None, translate=None, model="small"): def transcribe(path, language=None, translate=None):
#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"
@ -34,11 +33,10 @@ 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("-m", "--model", dest="model", default="small") parser.add_argument("-t", "--translate", dest="translate", default=None)
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, opts.model) result = transcribe(opts.path, opts.language, opts.translate)
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)

3
transcribe.py
View file

@ -1,4 +1,3 @@
import json
import logging import logging
import os import os
import shutil import shutil
@ -94,10 +93,12 @@ def prepare_joint_annotations(result, target_length=200):
def extract_subtitles(item, user, layer, translate, gpu=False, join_sentences=False): def extract_subtitles(item, user, layer, translate, gpu=False, join_sentences=False):
language = None language = None
"""
if "language" not in item.data: if "language" not in item.data:
language = None language = None
else: else:
language = ox.iso.langTo2Code(item.data["language"][0]) language = ox.iso.langTo2Code(item.data["language"][0])
"""
if not item.streams(): if not item.streams():
logger.error("skip item without media %s: %s", item.public_id) logger.error("skip item without media %s: %s", item.public_id)
return False return False