Merge branch 'main' of github.com:alibaba-damo-academy/FunASR add

游雁

2023-03-18 977519f54b6a4a96a919f47839baebdb2c38f9f9

Merge branch 'main' of github.com:alibaba-damo-academy/FunASR
add

 egs_modelscope/asr/paraformer/speech_paraformer_asr_nat-zh-cn-16k-common-vocab8404-online/infer.py

@@ -3,12 +3,17 @@
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks

from modelscope.utils.logger import get_logger
import logging
logger = get_logger(log_level=logging.CRITICAL)
logger.setLevel(logging.CRITICAL)

inference_pipeline = pipeline(
    task=Tasks.auto_speech_recognition,
    model='damo/speech_paraformer_asr_nat-zh-cn-16k-common-vocab8404-online',
    model_revision='v1.0.2')

waveform, sample_rate = torchaudio.load("asr_example_zh.wav")
waveform, sample_rate = torchaudio.load("waihu.wav")
speech_length = waveform.shape[1]
speech = waveform[0]

@@ -23,7 +28,7 @@
speech_cache = []
final_result = ""

while len(speech_buffer) > 0:
while len(speech_buffer) >= 960:
    if first_chunk:
        if len(speech_buffer) >= 14400:
            rec_result = inference_pipeline(audio_in=speech_buffer[0:14400], param_dict=param_dict)
@@ -42,11 +47,11 @@
            rec_result = inference_pipeline(audio_in=speech_buffer[:19200], param_dict=param_dict)
            speech_buffer = speech_buffer[9600:]
        else:
            cache_en["stride"] = len(speech_buffer) // 960
            cache_en["stride"] = len(speech_buffer) // 960 
            cache_en["pad_right"] = 0
            rec_result = inference_pipeline(audio_in=speech_buffer, param_dict=param_dict)
            speech_buffer = []
    if rec_result['text'] != "sil":
    if len(rec_result) !=0 and rec_result['text'] != "sil":
        final_result += rec_result['text']
    print(rec_result)
print(final_result)