Dev gzf deepspeed (#1732)

zhifu gao

2024-05-15 a0f03bd2a87d97d47a1636bbe6f0855a43160331

 examples/industrial_data_pretraining/emotion2vec/demo.py

@@ -6,12 +6,21 @@
from funasr import AutoModel

# model="iic/emotion2vec_base"
model = AutoModel(model="iic/emotion2vec_base_finetuned",
                  # vad_model="iic/speech_fsmn_vad_zh-cn-16k-common-pytorch",
                  # vad_model_revision="master",
                  # vad_kwargs={"max_single_segment_time": 2000},
                  )
# model="iic/emotion2vec_base_finetuned"
# model="iic/emotion2vec_plus_seed"
# model="iic/emotion2vec_plus_base"
model = "iic/emotion2vec_plus_large"

model = AutoModel(
    model=model,
    # vad_model="iic/speech_fsmn_vad_zh-cn-16k-common-pytorch",
    # vad_model_revision="master",
    # vad_kwargs={"max_single_segment_time": 2000},
)

wav_file = f"{model.model_path}/example/test.wav"
res = model.generate(wav_file, output_dir="./outputs", granularity="utterance", extract_embedding=False)
print(res)

res = model.generate(
    wav_file, output_dir="./outputs", granularity="utterance", extract_embedding=False
)
print(res)