ESPnet
multilingual
audio
codec
ftshijt commited on
Commit
f0bc82a
·
1 Parent(s): 0e4fe9b

Update model

Browse files
Files changed (29) hide show
  1. README.md +4 -7
  2. exp_bench_16k/codec_dac_16k_music_single/360epoch.pth +3 -0
  3. exp_bench_16k/codec_dac_16k_music_single/config.yaml +1 -1
  4. exp_bench_16k/codec_dac_16k_music_single/images/adv_loss.png +0 -0
  5. exp_bench_16k/codec_dac_16k_music_single/images/codec_commit_loss.png +0 -0
  6. exp_bench_16k/codec_dac_16k_music_single/images/codec_loss.png +0 -0
  7. exp_bench_16k/codec_dac_16k_music_single/images/codec_quantization_loss.png +0 -0
  8. exp_bench_16k/codec_dac_16k_music_single/images/discriminator_backward_time.png +0 -0
  9. exp_bench_16k/codec_dac_16k_music_single/images/discriminator_forward_time.png +0 -0
  10. exp_bench_16k/codec_dac_16k_music_single/images/discriminator_loss.png +0 -0
  11. exp_bench_16k/codec_dac_16k_music_single/images/discriminator_optim_step_time.png +0 -0
  12. exp_bench_16k/codec_dac_16k_music_single/images/discriminator_train_time.png +0 -0
  13. exp_bench_16k/codec_dac_16k_music_single/images/fake_loss.png +0 -0
  14. exp_bench_16k/codec_dac_16k_music_single/images/feat_match_loss.png +0 -0
  15. exp_bench_16k/codec_dac_16k_music_single/images/generator_backward_time.png +0 -0
  16. exp_bench_16k/codec_dac_16k_music_single/images/generator_forward_time.png +0 -0
  17. exp_bench_16k/codec_dac_16k_music_single/images/generator_optim_step_time.png +0 -0
  18. exp_bench_16k/codec_dac_16k_music_single/images/generator_train_time.png +0 -0
  19. exp_bench_16k/codec_dac_16k_music_single/images/gpu_max_cached_mem_GB.png +0 -0
  20. exp_bench_16k/codec_dac_16k_music_single/images/iter_time.png +0 -0
  21. exp_bench_16k/codec_dac_16k_music_single/images/loss.png +0 -0
  22. exp_bench_16k/codec_dac_16k_music_single/images/mel_loss.png +0 -0
  23. exp_bench_16k/codec_dac_16k_music_single/images/mel_loss_real.png +0 -0
  24. exp_bench_16k/codec_dac_16k_music_single/images/optim0_lr0.png +0 -0
  25. exp_bench_16k/codec_dac_16k_music_single/images/optim1_lr0.png +0 -0
  26. exp_bench_16k/codec_dac_16k_music_single/images/real_loss.png +0 -0
  27. exp_bench_16k/codec_dac_16k_music_single/images/reconstruct_loss.png +0 -0
  28. exp_bench_16k/codec_dac_16k_music_single/images/train_time.png +0 -0
  29. meta.yaml +3 -3
README.md CHANGED
@@ -35,14 +35,14 @@ cd egs2/amuse/codec1
35
  <details><summary>expand</summary>
36
 
37
  ```
38
- config: conf/train_dac_large_v1.4.yaml
39
  print_config: false
40
  log_level: INFO
41
  drop_last_iter: false
42
  dry_run: false
43
  iterator_type: chunk
44
  valid_iterator_type: null
45
- output_dir: exp_bench_16k/codec_dac_16k_music
46
  ngpu: 1
47
  seed: 777
48
  num_workers: 1
@@ -53,7 +53,7 @@ dist_world_size: 2
53
  dist_rank: 0
54
  local_rank: 0
55
  dist_master_addr: localhost
56
- dist_master_port: 60549
57
  dist_launcher: null
58
  multiprocessing_distributed: true
59
  unused_parameters: true
@@ -209,7 +209,7 @@ codec_conf:
209
  decoder_trim_right_ratio: 1.0
210
  decoder_final_activation: null
211
  decoder_final_activation_params: null
212
- quantizer_n_q: 8
213
  quantizer_bins: 1024
214
  quantizer_decay: 0.99
215
  quantizer_kmeans_init: true
@@ -217,9 +217,6 @@ codec_conf:
217
  quantizer_threshold_ema_dead_code: 2
218
  quantizer_target_bandwidth:
219
  - 0.5
220
- - 1
221
- - 2
222
- - 4
223
  quantizer_dropout: true
224
  sample_rate: 16000
225
  discriminator_params:
 
35
  <details><summary>expand</summary>
36
 
37
  ```
38
+ config: conf/train_dac_large_v1.4_single.yaml
39
  print_config: false
40
  log_level: INFO
41
  drop_last_iter: false
42
  dry_run: false
43
  iterator_type: chunk
44
  valid_iterator_type: null
45
+ output_dir: exp_bench_16k/codec_dac_16k_music_single
46
  ngpu: 1
47
  seed: 777
48
  num_workers: 1
 
53
  dist_rank: 0
54
  local_rank: 0
55
  dist_master_addr: localhost
56
+ dist_master_port: 53969
57
  dist_launcher: null
58
  multiprocessing_distributed: true
59
  unused_parameters: true
 
209
  decoder_trim_right_ratio: 1.0
210
  decoder_final_activation: null
211
  decoder_final_activation_params: null
212
+ quantizer_n_q: 1
213
  quantizer_bins: 1024
214
  quantizer_decay: 0.99
215
  quantizer_kmeans_init: true
 
217
  quantizer_threshold_ema_dead_code: 2
218
  quantizer_target_bandwidth:
219
  - 0.5
 
 
 
220
  quantizer_dropout: true
221
  sample_rate: 16000
222
  discriminator_params:
exp_bench_16k/codec_dac_16k_music_single/360epoch.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fb143c9984190dee1e066a4ff3d45134a253dbea07e2168c63388be9b2b99544
3
+ size 253700331
exp_bench_16k/codec_dac_16k_music_single/config.yaml CHANGED
@@ -16,7 +16,7 @@ dist_world_size: 2
16
  dist_rank: 0
17
  local_rank: 0
18
  dist_master_addr: localhost
19
- dist_master_port: 52201
20
  dist_launcher: null
21
  multiprocessing_distributed: true
22
  unused_parameters: true
 
16
  dist_rank: 0
17
  local_rank: 0
18
  dist_master_addr: localhost
19
+ dist_master_port: 53969
20
  dist_launcher: null
21
  multiprocessing_distributed: true
22
  unused_parameters: true
exp_bench_16k/codec_dac_16k_music_single/images/adv_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/codec_commit_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/codec_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/codec_quantization_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/discriminator_backward_time.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/discriminator_forward_time.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/discriminator_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/discriminator_optim_step_time.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/discriminator_train_time.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/fake_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/feat_match_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/generator_backward_time.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/generator_forward_time.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/generator_optim_step_time.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/generator_train_time.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/gpu_max_cached_mem_GB.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/iter_time.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/mel_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/mel_loss_real.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/optim0_lr0.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/optim1_lr0.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/real_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/reconstruct_loss.png CHANGED
exp_bench_16k/codec_dac_16k_music_single/images/train_time.png CHANGED
meta.yaml CHANGED
@@ -1,8 +1,8 @@
1
  espnet: '202402'
2
  files:
3
- model_file: exp_bench_16k/codec_dac_16k_music/360epoch.pth
4
  python: 3.10.13 | packaged by conda-forge | (main, Dec 23 2023, 15:26:55) [GCC 12.3.0]
5
- timestamp: 1736237745.352415
6
  torch: 2.6.0.dev20241209+cu124
7
  yaml_files:
8
- train_config: exp_bench_16k/codec_dac_16k_music/config.yaml
 
1
  espnet: '202402'
2
  files:
3
+ model_file: exp_bench_16k/codec_dac_16k_music_single/360epoch.pth
4
  python: 3.10.13 | packaged by conda-forge | (main, Dec 23 2023, 15:26:55) [GCC 12.3.0]
5
+ timestamp: 1736237840.855863
6
  torch: 2.6.0.dev20241209+cu124
7
  yaml_files:
8
+ train_config: exp_bench_16k/codec_dac_16k_music_single/config.yaml