| | |
|---|
| | | nj=32 |
|---|
| | | |
|---|
| | | inference_device="cuda" #"cpu" |
|---|
| | | inference_checkpoint="model.pt" |
|---|
| | | inference_checkpoint="model.pt.avg10" |
|---|
| | | inference_scp="wav.scp" |
|---|
| | | inference_batch_size=32 |
|---|
| | | |
|---|
| | |
|---|
| | | --config-path "${workspace}/conf" \ |
|---|
| | | --config-name "${config}" \ |
|---|
| | | ++train_data_set_list="${feats_dir}/data/${train_set}/audio_datasets.jsonl" \ |
|---|
| | | ++cmvn_file="${feats_dir}/data/${train_set}/cmvn.json" \ |
|---|
| | | ++dataset_conf.num_workers=$nj |
|---|
| | | ++cmvn_file="${feats_dir}/data/${train_set}/cmvn.json" |
|---|
| | | fi |
|---|
| | | |
|---|
| | | token_list=${feats_dir}/data/${lang}_token_list/$token_type/tokens.txt |
|---|
