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: 0x2620:3a680f3a1dbc94e0130ae67c92378c08e8fc2bb5
Branches Tags
0x2620:main

2 commits
98c680e89c ... 3a680f3a1d

Author SHA1 Message Date
j
3a680f3a1d expose model to manage command 2025-01-24 12:10:35 +05:30
j
1bfa67ad5d pass model 2025-01-24 11:52:29 +05:30
2 changed files with 7 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

2
management/commands/transcribe.py
View file

@ -21,6 +21,8 @@ class Command(BaseCommand):
default=None, help='import into layer (default subtitle layer)') default=None, help='import into layer (default subtitle layer)')
parser.add_argument('--translate', type=str, dest='translate', parser.add_argument('--translate', type=str, dest='translate',
default='', help='list of languages to translate: (i.e. hi:en,de:en') default='', help='list of languages to translate: (i.e. hi:en,de:en')
parser.add_argument('--model', type=str, dest='model',
default='small', help='whisper model (small, large-v3-turbo, default: small)')
parser.add_argument('--gpu', action='store_true', dest='gpu', parser.add_argument('--gpu', action='store_true', dest='gpu',
default=False, help='user GPU (default: disabled)') default=False, help='user GPU (default: disabled)')
parser.add_argument('--join-sentences', action='store_true', dest='join_sentences', parser.add_argument('--join-sentences', action='store_true', dest='join_sentences',

8
transcribe.py
View file

@ -92,7 +92,7 @@ def prepare_joint_annotations(result, target_length=200):
return annotations return annotations
def extract_subtitles(item, user, layer, translate, gpu=False, join_sentences=False): def extract_subtitles(item, user, layer, translate, gpu=False, join_sentences=False, model="small"):
language = None language = None
if "language" not in item.data: if "language" not in item.data:
language = None language = None
@ -108,6 +108,7 @@ def extract_subtitles(item, user, layer, translate, gpu=False, join_sentences=Fa
run_py = os.path.join(os.path.dirname(os.path.abspath(__file__)), "run_whisper.py") run_py = os.path.join(os.path.dirname(os.path.abspath(__file__)), "run_whisper.py")
cmd = ["/opt/whisper-timestamped/bin/python", run_py] cmd = ["/opt/whisper-timestamped/bin/python", run_py]
cmd += ["--model", model]
if language: if language:
cmd += ["--language", language] cmd += ["--language", language]
if translate and language in translate: if translate and language in translate:
@ -148,7 +149,7 @@ def extract_subtitles(item, user, layer, translate, gpu=False, join_sentences=Fa
return True return True
def extract_subtitles_cmd(item, user, layer, translate, gpu=False): def extract_subtitles_cmd(item, user, layer, translate, gpu=False, model="small"):
if "language" not in item.data: if "language" not in item.data:
language = None language = None
else: else:
@ -159,7 +160,7 @@ def extract_subtitles_cmd(item, user, layer, translate, gpu=False):
src = item.streams()[0].media.path src = item.streams()[0].media.path
tmp = tempfile.mkdtemp() tmp = tempfile.mkdtemp()
cmd = ["/opt/whisper-timestamped/bin/whisper_timestamped", "--model", "turbo"] cmd = ["/opt/whisper-timestamped/bin/whisper_timestamped", "--model", model]
if language: if language:
cmd += ["--language", language] cmd += ["--language", language]
if translate and language in translate: if translate and language in translate:
@ -240,6 +241,7 @@ def main(**kwargs):
translate, translate,
kwargs.get("gpu"), kwargs.get("gpu"),
join_sentences=kwargs.get("join_sentences"), join_sentences=kwargs.get("join_sentences"),
model=kwargs.get("model", "small")
): ):
done.items.add(item) done.items.add(item)
queue.items.remove(item) queue.items.remove(item)