update

examples/nx_mpnet/run.sh

@@ -3,24 +3,10 @@
 : <<'END'
 
 
-sh run.sh --stage 2 --stop_stage 2 --system_version windows --file_folder_name file_dir --final_model_name mpnet-aishell-20250224 \
---noise_dir "E:/Users/tianx/HuggingDatasets/nx_noise/data/noise" \
---speech_dir "E:/programmer/asr_datasets/aishell/data_aishell/wav/train"
-
-
-sh run.sh --stage 3 --stop_stage 3 --system_version centos --file_folder_name file_dir --final_model_name mpnet-aishell-20250224 \
---noise_dir "/data/tianxing/HuggingDatasets/nx_noise/data/noise" \
---speech_dir "/data/tianxing/HuggingDatasets/aishell/data_aishell/wav/train"
-
-sh run.sh --stage 5 --stop_stage 5 --system_version centos --file_folder_name file_dir --final_model_name mpnet-aishell-20250224 \
---noise_dir "/data/tianxing/HuggingDatasets/nx_noise/data/noise" \
---speech_dir "/data/tianxing/HuggingDatasets/aishell/data_aishell/wav/train"
-
-
-sh run.sh --stage 1 --stop_stage 2 --system_version centos --file_folder_name file_dir --final_model_name mpnet-nx-speech-20250224 \
+sh run.sh --stage 1 --stop_stage 2 --system_version centos --file_folder_name file_dir --final_model_name nx-mpnet-aishell-20250224 \
 --noise_dir "/data/tianxing/HuggingDatasets/nx_noise/data/noise" \
---speech_dir "/data/tianxing/HuggingDatasets/nx_noise/data/speech" \
---max_epochs 1
+--speech_dir "/data/tianxing/HuggingDatasets/aishell/data_aishell/wav/train" \
+--max_epochs 100
 
 
 END

examples/nx_mpnet/yaml/config.yaml

@@ -1,27 +1,38 @@
-model_name: "mpnet"
-
-num_gpus: 0
-batch_size: 3
-learning_rate: 0.0005
-adam_b1: 0.8
-adam_b2: 0.99
-lr_decay: 0.99
-seed: 1234
-
-dense_channel: 64
-compress_factor: 0.3
-num_tsconformers: 4
-beta: 2.0
+model_name: "nx_denoise"
 
 sample_rate: 8000
 segment_size: 16000
 n_fft: 512
-hop_size: 80
 win_size: 200
+hop_size: 80
+
+dense_num_blocks: 4
+dense_hidden_size: 64
 
-num_workers: 4
+mask_num_blocks: 4
+mask_hidden_size: 64
 
-dist_config:
-  dist_backend: nccl
-  dist_url: tcp://localhost:54321
-  world_size: 1
+phase_num_blocks: 4
+phase_hidden_size: 64
+
+tsfm_hidden_size: 64
+tsfm_attention_heads: 4
+tsfm_num_blocks: 4
+tsfm_dropout_rate: 0.0
+tsfm_max_time_relative_position: 1024
+tsfm_max_freq_relative_position: 128
+tsfm_chunk_size: 1
+tsfm_num_left_chunks: 64
+tsfm_num_right_chunks: 2
+
+discriminator_dim: 32
+discriminator_in_channel: 2
+
+compress_factor: 0.3
+
+batch_size: 4
+learning_rate: 0.0005
+adam_b1: 0.8
+adam_b2: 0.99
+lr_decay: 0.99
+seed: 1234

toolbox/torchaudio/models/nx_mpnet/yaml/config.yaml

@@ -0,0 +1,38 @@
+model_name: "nx_denoise"
+
+sample_rate: 8000
+segment_size: 16000
+n_fft: 512
+win_size: 200
+hop_size: 80
+
+dense_num_blocks: 4
+dense_hidden_size: 64
+
+mask_num_blocks: 4
+mask_hidden_size: 64
+
+phase_num_blocks: 4
+phase_hidden_size: 64
+
+tsfm_hidden_size: 64
+tsfm_attention_heads: 4
+tsfm_num_blocks: 4
+tsfm_dropout_rate: 0.0
+tsfm_max_time_relative_position: 1024
+tsfm_max_freq_relative_position: 128
+tsfm_chunk_size: 1
+tsfm_num_left_chunks: 64
+tsfm_num_right_chunks: 2
+
+discriminator_dim: 32
+discriminator_in_channel: 2
+
+compress_factor: 0.3
+
+batch_size: 4
+learning_rate: 0.0005
+adam_b1: 0.8
+adam_b2: 0.99
+lr_decay: 0.99
+seed: 1234