Can SenseVoice works with speaker diarization model like cam++ ?

[shenxiangzhuang]

trafficstars

With demo code, I got error like "can not get timestamp":

from funasr import AutoModel
from funasr.utils.postprocess_utils import rich_transcription_postprocess

model_dir = "iic/SenseVoiceSmall"

model = AutoModel(
    model=model_dir,
    vad_model="fsmn-vad",
    vad_kwargs={"max_single_segment_time": 30000},
    device="cuda:0",
)

# en
res = model.generate(
    input=f"{model.model_path}/example/en.mp3",
    cache={},
    language="auto",  # "zn", "en", "yue", "ja", "ko", "nospeech"
    use_itn=True,
    batch_size_s=60,
    merge_vad=True,  #
    merge_length_s=15,
)
text = rich_transcription_postprocess(res[0]["text"])
print(text)

Then I tried to set output_timestamp=True, the code like:

self.model = AutoModel(
    model="iic/SenseVoiceSmall",
    output_timestamp=True,
    # vad_model="fsmn-vad",
    # vad_kwargs={"max_single_segment_time": 30000},
    vad_model="fsmn-vad",
    # vad_kwargs={"max_single_segment_time": 180000, "merge_vad": True, "merge_length_s": 15},
    punc_model="ct-punc-c", punc_model_revision="v2.0.4",
    spk_model="cam++", spk_model_revision="v2.0.2",
    device="cuda:0",
)

But I got another error:

  File ".../python3.10/site-packages/funasr/models/campplus/utils.py", line 203, in distribute_spk
    overlap = max(min(sentence_end, spk_ed) - max(sentence_start, spk_st), 0)
TypeError: '>' not supported between instances of 'float' and 'NoneType'

So my question is can SenseVoice works with speaker diarization model like cam++ or my setting should be changed to make it work?