admin/scribe
1
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity

solved path issues

Browse Source
This commit is contained in:
Jaikinator
2023-06-30 18:41:43 +02:00
parent 11fce3abef
commit cd35ad8903
1 changed files with 23 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files
autotranscript/misc.py
+23 -26
View File
@@ -1,4 +1,3 @@
from pyannote.audio import Pipeline from pyannote.audio import Pipeline
from whisper import Whisper, load_model from whisper import Whisper, load_model
import os import os
@@ -6,15 +5,18 @@ import glob
from warnings import warn from warnings import warn
import yaml import yaml
WHISPER_DEFAULT_PATH = os.path.relpath(os.path.join(os.path.dirname(__file__), CACHE_DIR = os.getenv(
"models", "whisper")) "AUTOT_CACHE",
os.path.expanduser("~/.cache/torch/models"),
)
PYANNOTE_DEFAULT_PATH = os.path.relpath(os.path.join(os.path.dirname(__file__), WHISPER_DEFAULT_PATH = os.path.join(CACHE_DIR, "whisper")
"models", "pyannote",
"speaker_diarization", "config.yaml"))
PYANNOTE_DEFAULT_PATH = os.path.join(CACHE_DIR, "pyannote")
def config_diarization_yaml(file, path_to_segmentation = None, path_to_embedding = None): PYANNOTE_DEFAULT_CONFIG = os.path.join(PYANNOTE_DEFAULT_PATH, "config.yaml")
def config_diarization_yaml(file, path_to_segmentation = None):
""" """
Configure diarization pipeline from yaml file to use the model offline Configure diarization pipeline from yaml file to use the model offline
and avoid manuel file manipulation. and avoid manuel file manipulation.
@@ -28,30 +30,25 @@ def config_diarization_yaml(file, path_to_segmentation = None, path_to_embedding
if path_to_segmentation: if path_to_segmentation:
yml["pipeline"]["params"]["segmentation"] = path_to_segmentation yml["pipeline"]["params"]["segmentation"] = path_to_segmentation
else: else:
yml["pipeline"]["params"]["segmentation"] = os.path.relpath(os.path.join( yml["pipeline"]["params"]["segmentation"] = os.path.join(PYANNOTE_DEFAULT_PATH, "pytorch_model.bin")
os.path.dirname(__file__),
"models", "pyannote",
"segmentation",
"pytorch_model.bin"))
if path_to_embedding: # if path_to_embedding:
yml["pipeline"]["params"]["embedding"] = path_to_embedding # yml["pipeline"]["params"]["embedding"] = path_to_embedding
else: # else:
yml["pipeline"]["params"]["embedding"] = os.path.relpath( # yml["pipeline"]["params"]["embedding"] = os.path.relpath(
os.path.join( # os.path.join(
os.path.dirname(__file__), # os.path.dirname(__file__),
"models", "pyannote", # "models", "pyannote",
"speechbrain", # "speechbrain",
"spkrec-ecapa-voxceleb", # "spkrec-ecapa-voxceleb",
"embedding_model.ckpt")) # "embedding_model.ckpt"))
if not os.path.exists(yml["pipeline"]["params"]["segmentation"]): if not os.path.exists(yml["pipeline"]["params"]["segmentation"]):
raise FileNotFoundError(f"Segmentation model not found at {yml['pipeline']['params']['segmentation']}") raise FileNotFoundError(f"Segmentation model not found at {yml['pipeline']['params']['segmentation']}")
if not os.path.exists(yml["pipeline"]["params"]["embedding"]): # if not os.path.exists(yml["pipeline"]["params"]["embedding"]):
raise FileNotFoundError(f"Embedding model not found at {yml['pipeline']['params']['embedding']}") # raise FileNotFoundError(f"Embedding model not found at {yml['pipeline']['params']['embedding']}")
with open(file, "w") as stream: with open(file, "w") as stream:
yaml.dump(yml, stream) yaml.dump(yml, stream)
stream.close() stream.close()