add mossformer inference recipe
| New file |
| | |
|---|
| | | import numpy |
|---|
| | | import soundfile as sf |
|---|
| | | from modelscope.pipelines import pipeline |
|---|
| | | from modelscope.utils.constant import Tasks |
|---|
| | | |
|---|
| | | input = 'https://modelscope.cn/api/v1/models/damo/speech_separation_mossformer_8k_pytorch/repo?Revision=master&FilePath=examples/mix_speech1.wav' |
|---|
| | | separation = pipeline( |
|---|
| | | Tasks.funasr_speech_separation, |
|---|
| | | model='damo/speech_separation_mossformer_8k_pytorch', |
|---|
| | | output_dir='./', |
|---|
| | | model_revision='v1.0.1') |
|---|
| | | result = separation(audio_in=input) |
|---|
| | | for i, signal in enumerate(result): |
|---|
| | | save_file = f'output_spk_{i+1}.wav' |
|---|
| | | sf.write(save_file, signal, 8000) |
|---|
