| | |
| | | quantize_dtype: Optional[str] = "float16", |
| | | streaming: Optional[bool] = False, |
| | | simu_streaming: Optional[bool] = False, |
| | | full_utt: Optional[bool] = False, |
| | | chunk_size: Optional[int] = 16, |
| | | left_context: Optional[int] = 16, |
| | | right_context: Optional[int] = 0, |
| | |
| | | quantize_dtype=quantize_dtype, |
| | | streaming=streaming, |
| | | simu_streaming=simu_streaming, |
| | | full_utt=full_utt, |
| | | chunk_size=chunk_size, |
| | | left_context=left_context, |
| | | right_context=right_context, |
| | |
| | | _end = (i + 1) * speech2text._ctx |
| | | |
| | | speech2text.streaming_decode( |
| | | speech[i * speech2text._ctx: _end], is_final=False |
| | | speech[i * speech2text._ctx: _end + speech2text._right_ctx], is_final=False |
| | | ) |
| | | |
| | | final_hyps = speech2text.streaming_decode( |
| | |
| | | ) |
| | | elif speech2text.simu_streaming: |
| | | final_hyps = speech2text.simu_streaming_decode(**batch) |
| | | elif speech2text.full_utt: |
| | | final_hyps = speech2text.full_utt_decode(**batch) |
| | | else: |
| | | final_hyps = speech2text(**batch) |
| | | |
| | |
| | | group.add_argument("--ngram_weight", type=float, default=0.9, help="ngram weight") |
| | | group.add_argument("--streaming", type=str2bool, default=False) |
| | | group.add_argument("--simu_streaming", type=str2bool, default=False) |
| | | group.add_argument("--full_utt", type=str2bool, default=False) |
| | | group.add_argument("--chunk_size", type=int, default=16) |
| | | group.add_argument("--left_context", type=int, default=16) |
| | | group.add_argument("--right_context", type=int, default=0) |