pyannote-audio
pyannote-audio copied to clipboard
pyannote
Offline loading of pipeline ('NoneType' object has no attribute 'eval')
I am not sure if i am missing something. I followed the documentation in how to load a pipeline for speaker diarization offline.
i followed this description: https://github.com/pyannote/pyannote-audio/blob/develop/tutorials/applying_a_pipeline.ipynb
But used the config.yaml from https://huggingface.co/pyannote/speaker-diarization instead of the VAD that is used in the offline use section since i want to use speaker diarization and not voice activity detection. (thats a bit confusing since at the top of that description, speaker diarization is used.)
i try to load it like this:
from pyannote.audio import Pipeline as PyannotePipeline
PyannotePipeline.from_pretrained(str(Path(cache_pyannote_path / "speaker-diarization" / "pipeline_config.yaml").resolve()))
but i get the following error:
2023-04-07 20:39:16 - +--------------------- Traceback (most recent call last) ---------------------+
2023-04-07 20:39:16 - | E:\AI\xyz\xyz.py:566 |
2023-04-07 20:39:16 - | in <module> |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:16 - | 563 return bool(string) |
2023-04-07 20:39:16 - | 564 |
2023-04-07 20:39:16 - | 565 |
2023-04-07 20:39:16 - | > 566 main() |
2023-04-07 20:39:16 - | 567 |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:16 - | E:\Python\Python310\lib\site-packages\click\core.py:1130 in __call__ |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:16 - | 1127 |
2023-04-07 20:39:16 - | 1128 def __call__(self, *args: t.Any, **kwargs: t.Any) -> t.Any: |
2023-04-07 20:39:16 - | 1129 """Alias for :meth:`main`.""" |
2023-04-07 20:39:16 - | > 1130 return self.main(*args, **kwargs) |
2023-04-07 20:39:16 - | 1131 |
2023-04-07 20:39:16 - | 1132 |
2023-04-07 20:39:16 - | 1133 class Command(BaseCommand): |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:16 - | E:\Python\Python310\lib\site-packages\click\core.py:1055 in main |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:16 - | 1052 try: |
2023-04-07 20:39:16 - | 1053 try: |
2023-04-07 20:39:16 - | 1054 with self.make_context(prog_name, args, **extra) as |
2023-04-07 20:39:16 - | > 1055 rv = self.invoke(ctx) |
2023-04-07 20:39:16 - | 1056 if not standalone_mode: |
2023-04-07 20:39:16 - | 1057 return rv |
2023-04-07 20:39:16 - | 1058 # it's not safe to `ctx.exit(rv)` here! |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:16 - | E:\Python\Python310\lib\site-packages\click\core.py:1404 in invoke |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:16 - | 1401 echo(style(message, fg="red"), err=True) |
2023-04-07 20:39:16 - | 1402 |
2023-04-07 20:39:16 - | 1403 if self.callback is not None: |
2023-04-07 20:39:16 - | > 1404 return ctx.invoke(self.callback, **ctx.params) |
2023-04-07 20:39:16 - | 1405 |
2023-04-07 20:39:16 - | 1406 def shell_complete(self, ctx: Context, incomplete: str) -> t.Lis |
2023-04-07 20:39:16 - | 1407 """Return a list of completions for the incomplete value. Lo |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:16 - | E:\Python\Python310\lib\site-packages\click\core.py:760 in invoke |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:16 - | 757 |
2023-04-07 20:39:16 - | 758 with augment_usage_errors(__self): |
2023-04-07 20:39:16 - | 759 with ctx: |
2023-04-07 20:39:16 - | > 760 return __callback(*args, **kwargs) |
2023-04-07 20:39:16 - | 761 |
2023-04-07 20:39:16 - | 762 def forward( |
2023-04-07 20:39:16 - | 763 __self, __cmd: "Command", *args: t.Any, **kwargs: t.Any # n |
2023-04-07 20:39:16 - | |
2023-04-07 20:39:18 - | E:\Python\Python310\lib\site-packages\click\decorators.py:26 in new_func |
2023-04-07 20:39:18 - | |
2023-04-07 20:39:18 - | 23 """ |
2023-04-07 20:39:18 - | 24 |
2023-04-07 20:39:18 - | 25 def new_func(*args, **kwargs): # type: ignore |
2023-04-07 20:39:18 - | > 26 return f(get_current_context(), *args, **kwargs) |
2023-04-07 20:39:18 - | 27 |
2023-04-07 20:39:18 - | 28 return update_wrapper(t.cast(F, new_func), f) |
2023-04-07 20:39:18 - | 29 |
2023-04-07 20:39:18 - | |
2023-04-07 20:39:18 - | E:\AI\xyz\xyz.py:373 |
2023-04-07 20:39:18 - | in main |
2023-04-07 20:39:18 - | |
2023-04-07 20:39:18 - | 370 |
2023-04-07 20:39:18 - | 371 # load speaker diarization model |
2023-04-07 20:39:18 - | 372 #diarization_model = PyannoteModel.from_pretrained(str(Path(c |
2023-04-07 20:39:18 - | > 373 diarization_pipeline = PyannotePipeline.from_pretrained(str(P |
2023-04-07 20:39:18 - | 374 |
2023-04-07 20:39:18 - | 375 # num_samples = 1536 |
2023-04-07 20:39:18 - | 376 num_samples = int(settings.SetOption("vad_num_samples", |
2023-04-07 20:39:18 - | |
2023-04-07 20:39:18 - | E:\Python\Python310\lib\site-packages\pyannote\audio\core\pipeline.py:126 |
2023-04-07 20:39:18 - | in from_pretrained |
2023-04-07 20:39:18 - | |
2023-04-07 20:39:18 - | 123 ) |
2023-04-07 20:39:18 - | 124 params = config["pipeline"].get("params", {}) |
2023-04-07 20:39:18 - | 125 params.setdefault("use_auth_token", use_auth_token) |
2023-04-07 20:39:18 - | > 126 pipeline = Klass(**params) |
2023-04-07 20:39:18 - | 127 |
2023-04-07 20:39:18 - | 128 # freeze parameters |
2023-04-07 20:39:18 - | 129 if "freeze" in config: |
2023-04-07 20:39:18 - | |
2023-04-07 20:39:18 - | E:\Python\Python310\lib\site-packages\pyannote\audio\pipelines\speaker_diar |
2023-04-07 20:39:18 - | ization.py:125 in __init__ |
2023-04-07 20:39:18 - | |
2023-04-07 20:39:18 - | 122 super().__init__() |
2023-04-07 20:39:18 - | 123 |
2023-04-07 20:39:18 - | 124 self.segmentation_model = segmentation |
2023-04-07 20:39:18 - | > 125 model: Model = get_model(segmentation, use_auth_token=use_aut |
2023-04-07 20:39:18 - | 126 |
2023-04-07 20:39:18 - | 127 self.segmentation_batch_size = segmentation_batch_size |
2023-04-07 20:39:18 - | 128 self.segmentation_duration = ( |
2023-04-07 20:39:18 - | |
2023-04-07 20:39:18 - | E:\Python\Python310\lib\site-packages\pyannote\audio\pipelines\utils\getter |
2023-04-07 20:39:18 - | .py:89 in get_model |
2023-04-07 20:39:18 - | |
2023-04-07 20:39:18 - | 86 f"expected `str` or `dict`." |
2023-04-07 20:39:18 - | 87 ) |
2023-04-07 20:39:18 - | 88 |
2023-04-07 20:39:18 - | > 89 model.eval() |
2023-04-07 20:39:18 - | 90 return model |
2023-04-07 20:39:18 - | 91 |
2023-04-07 20:39:18 - | 92 |
2023-04-07 20:39:18 - +-----------------------------------------------------------------------------+
2023-04-07 20:39:18 - AttributeError: 'NoneType' object has no attribute 'eval'
Thank you for your issue. Give us a little time to review it.
PS. You might want to check the FAQ if you haven't done so already.
This is an automated reply, generated by FAQtory
![github-actions[bot] avatar](https://avatars.githubusercontent.com/in/15368?v=4 "Published by a pub.dev verified publisher"){kind=small}
- Visit hf.co/pyannote/speaker-diarization
- hf.co/pyannote/segmentation
Accept user conditions of Both Models
And add the user token during downloading the model from pre-trianed
from pyannote.audio import Pipeline
pipeline = Pipeline.from_pretrained("pyannote/speaker-diarization",
use_auth_token="TOKEN HERE")
Visit hf.co/pyannote/speaker-diarization
- hf.co/pyannote/segmentation
Accept user conditions of Both Models
And add the user token during downloading the model from pre-trianed
from pyannote.audio import Pipeline pipeline = Pipeline.from_pretrained("pyannote/speaker-diarization", use_auth_token="TOKEN HERE")
The whole reason to do offline loading is that you do not need a token for every user of the software. I can't expect that every user will create a huggingface account etc.
- Edit
your/path/to/pyannote/speaker-diarization/config.yaml
pipeline:
name: pyannote.audio.pipelines.SpeakerDiarization
params:
clustering: AgglomerativeClustering
embedding: your/path/to/speechbrain/spkrec-ecapa-voxceleb # Folder, must contains `speechbrain` keyword.
embedding_batch_size: 32
embedding_exclude_overlap: true
segmentation: your/path/to/pyannote/segmentation/[email protected] # File
segmentation_batch_size: 32
params:
clustering:
method: centroid
min_cluster_size: 15
threshold: 0.7153814381597874
segmentation:
min_duration_off: 0.5817029604921046
threshold: 0.4442333667381752
- Edit
pyannote/audio/pipelines/speaker_verification.py(version 2.1.1)
self.classifier_ = SpeechBrain_EncoderClassifier.from_hparams(
source=self.embedding,
savedir=self.embedding if Path(self.embedding).exists() else f"{CACHE_DIR}/speechbrain",
run_opts={"device": self.device},
use_auth_token=use_auth_token,
)
- Speaker diarization
from pyannote.audio import Pipeline
pipeline = Pipeline.from_pretrained("your/path/to/pyannote/speaker-diarization/config.yaml")
thanks. but still no luck. I placed the spkrec-ecapa-voxceleb stuff besides the pipeline config and changed the config accordingly:
pipeline:
name: pyannote.audio.pipelines.SpeakerDiarization
params:
clustering: AgglomerativeClustering
embedding: speechbrain/spkrec-ecapa-voxceleb
embedding_batch_size: 32
embedding_exclude_overlap: true
segmentation: pytorch_model.bin
segmentation_batch_size: 32
params:
clustering:
method: centroid
min_cluster_size: 15
threshold: 0.7153814381597874
segmentation:
min_duration_off: 0.5817029604921046
threshold: 0.4442333667381752
and changed the speaker_verification.py like you mentioned.
You need to download from speechbrain/spkrec-ecapa-voxceleb to the local speechbrain/spkrec-ecapa-voxceleb directory.
classifier.ckpt embedding_model.ckpt hyperparams.yaml label_encoder.ckpt mean_var_norm_emb.ckpt
Hello @hbredin ,
Would a PR containing the following point be accepted for offline model loading ?
2. Edit `pyannote/audio/pipelines/speaker_verification.py`(version 2.1.1)self.classifier_ = SpeechBrain_EncoderClassifier.from_hparams( source=self.embedding, savedir=self.embedding if Path(self.embedding).exists() else f"{CACHE_DIR}/speechbrain", run_opts={"device": self.device}, use_auth_token=use_auth_token, )
The modified line is savedir=self.embedding if Path(self.embedding).exists() else f"{CACHE_DIR}/speechbrain", at https://github.com/pyannote/pyannote-audio/blob/11b56a137a578db9335efc00298f6ec1932e6317/pyannote/audio/pipelines/speaker_verification.py#L260
I'd gladly have a look at a PR facilitating the offline use of pyannote. Would be nice to also update the related part of the documentation.
The tutorial doesn't work? Because I'm getting the same error when running it:
from pyannote.audio import Pipeline
pipeline = Pipeline.from_pretrained('pyannote/speaker-diarization', use_auth_token=True)
Error:
model.eval()
return model
AttributeError: 'NoneType' object has no attribute 'eval'
Hi @hbredin
I am also facing similar issue with offline usage with VAD config.
pyannote.audio version == 2.1.1
Config
pipeline: name: pyannote.audio.pipelines.VoiceActivityDetection params: segmentation: pytorch_model.bin params: min_duration_off: 0.09791355693027545 min_duration_on: 0.05537587440407595 offset: 0.4806866463041527 onset: 0.8104268538848918
Code pipeline = Pipeline.from_pretrained(f"vad_config.yaml")
Error: AttributeError: 'NoneType' object has no attribute 'eval'
Hello. Are there plans to make the offline use of speaker-diarization-3.0 pipeline work? I tried above suggestions to no avail.
pyannote models and pipelines have always been usable offline.
The documentation is just... missing.
- download the segmentation model
- download the embedding model
- copy the parameters of the pipeline configuration file
- edit it to point to the local models
Also, feel free to make a PR improving the documentation!
except that this is exactly what i tried some time ago without success.
Would have to try it again to see if it works now or if i was just missing something else in detail.
So yes. an updated documentation would help a lot if someone gets this to work and would update it.
issue is still there as of today, the model is not found for some reason and a none value is returned, can someone look into this please.
I did not see this issue when proposing the respective PR: https://github.com/pyannote/pyannote-audio/pull/1682
please check if the new tutorial addresses these issues: https://github.com/pyannote/pyannote-audio/blob/develop/tutorials/community/offline_usage_speaker_diarization.ipynb
- Edit
your/path/to/pyannote/speaker-diarization/config.yamlpipeline: name: pyannote.audio.pipelines.SpeakerDiarization params: clustering: AgglomerativeClustering embedding: your/path/to/speechbrain/spkrec-ecapa-voxceleb # Folder, must contains `speechbrain` keyword. embedding_batch_size: 32 embedding_exclude_overlap: true segmentation: your/path/to/pyannote/segmentation/[email protected] # File segmentation_batch_size: 32 params: clustering: method: centroid min_cluster_size: 15 threshold: 0.7153814381597874 segmentation: min_duration_off: 0.5817029604921046 threshold: 0.4442333667381752
- Edit
pyannote/audio/pipelines/speaker_verification.py(version 2.1.1)self.classifier_ = SpeechBrain_EncoderClassifier.from_hparams( source=self.embedding, savedir=self.embedding if Path(self.embedding).exists() else f"{CACHE_DIR}/speechbrain", run_opts={"device": self.device}, use_auth_token=use_auth_token, )
- Speaker diarization
from pyannote.audio import Pipeline pipeline = Pipeline.from_pretrained("your/path/to/pyannote/speaker-diarization/config.yaml")
按照你的方法，还是报错huggingface_hub.utils._validators.HFValidationError: Repo id must be in the form 'repo_name' or 'namespace/repo_name'
