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

游雁

2024-02-21 741d089eb9dd9be7b6e2cabbd40fc0a784eb38f3

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

 examples/industrial_data_pretraining/seaco_paraformer/demo.py

@@ -11,15 +11,16 @@
                  vad_model_revision="v2.0.4",
                  punc_model="damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch",
                  punc_model_revision="v2.0.4",
                  spk_model="damo/speech_campplus_sv_zh-cn_16k-common",
                  spk_model_revision="v2.0.2",
                  # spk_model="damo/speech_campplus_sv_zh-cn_16k-common",
                  # spk_model_revision="v2.0.2",
                  )


# example1
res = model.generate(input="https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav",
                     hotword='达摩院 魔搭',
                     # preset_spk_num=2,
                     # return_raw_text=True,     # return raw text recognition results splited by space of equal length with timestamp
                     # preset_spk_num=2,         # preset speaker num for speaker cluster model
                     # sentence_timestamp=True,  # return sentence level information when spk_model is not given
                    )
print(res)

 funasr/auto/auto_model.py

@@ -380,11 +380,13 @@
                        else:
                            result[k] += restored_data[j][k]
                            
            return_raw_text = kwargs.get('return_raw_text', False)            
            # step.3 compute punc model
            if self.punc_model is not None:
                self.punc_kwargs.update(cfg)
                punc_res = self.inference(result["text"], model=self.punc_model, kwargs=self.punc_kwargs, disable_pbar=True, **cfg)
                raw_text = copy.copy(result["text"])
                if return_raw_text: result['raw_text'] = raw_text
                result["text"] = punc_res[0]["text"]
            else:
                raw_text = None
@@ -405,7 +407,7 @@
                            logging.error("Only 'iic/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch' \
                                and 'iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch'\
                                can predict timestamp, and speaker diarization relies on timestamps.")
                        sentence_list.append({"start": vadsegment[0],\
                        sentence_list.append({"start": vadsegment[0],
                                                "end": vadsegment[1],
                                                "sentence": res['text'],
                                                "timestamp": res['timestamp']})
@@ -414,15 +416,17 @@
                        logging.error("Only 'iic/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch' \
                            and 'iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch'\
                            can predict timestamp, and speaker diarization relies on timestamps.")
                    sentence_list = timestamp_sentence(punc_res[0]['punc_array'], \
                                                        result['timestamp'], \
                                                        raw_text)
                    sentence_list = timestamp_sentence(punc_res[0]['punc_array'],
                                                       result['timestamp'],
                                                       raw_text,
                                                       return_raw_text=return_raw_text)
                distribute_spk(sentence_list, sv_output)
                result['sentence_info'] = sentence_list
            elif kwargs.get("sentence_timestamp", False):
                sentence_list = timestamp_sentence(punc_res[0]['punc_array'], \
                                                        result['timestamp'], \
                                                        raw_text)
                sentence_list = timestamp_sentence(punc_res[0]['punc_array'],
                                                   result['timestamp'],
                                                   raw_text,
                                                   return_raw_text=return_raw_text)
                result['sentence_info'] = sentence_list
            if "spk_embedding" in result: del result['spk_embedding']
                    

 funasr/models/paraformer/model.py

@@ -537,7 +537,6 @@
                    
                    result_i = {"key": key[i], "text": text_postprocessed}

                    
                    if ibest_writer is not None:
                        ibest_writer["token"][key[i]] = " ".join(token)
                        # ibest_writer["text"][key[i]] = text

 funasr/utils/timestamp_tools.py

@@ -98,7 +98,7 @@
    return res_txt, res


def timestamp_sentence(punc_id_list, timestamp_postprocessed, text_postprocessed):
def timestamp_sentence(punc_id_list, timestamp_postprocessed, text_postprocessed, return_raw_text=False):
    punc_list = ['，', '。', '？', '、']
    res = []
    if text_postprocessed is None:
@@ -142,14 +142,23 @@

        punc_id = int(punc_id) if punc_id is not None else 1
        sentence_end = timestamp[1] if timestamp is not None else sentence_end

        sentence_text_seg = sentence_text_seg[:-1] if sentence_text_seg[-1] == ' ' else sentence_text_seg
        if punc_id > 1:
            sentence_text += punc_list[punc_id - 2]
            if return_raw_text:
            res.append({
                'text': sentence_text,
                "start": sentence_start,
                "end": sentence_end,
                "timestamp": ts_list
                    "timestamp": ts_list,
                    'raw_text': sentence_text_seg,
                })
            else:
                res.append({
                    'text': sentence_text,
                    "start": sentence_start,
                    "end": sentence_end,
                    "timestamp": ts_list,
            })
            sentence_text = ''
            sentence_text_seg = ''