should be ok on the h100

tts/StyleTTS2Train/Models/LJSpeech/config_harness.yml

@@ -0,0 +1,110 @@
+log_dir: "Models/LJSpeech"
+save_freq: 5
+log_interval: 10
+device: "cuda"
+epochs: 50 # number of finetuning epoch (1 hour of data)
+batch_size: 8
+max_len: 400 # maximum number of frames
+pretrained_model: "Models/LibriTTS/epochs_2nd_00020.pth"
+second_stage_load_pretrained: true # set to true if the pre-trained model is for 2nd stage
+load_only_params: true # set to true if do not want to load epoch numbers and optimizer parameters
+
+F0_path: "Utils/JDC/bst.t7"
+ASR_config: "Utils/ASR/config.yml"
+ASR_path: "Utils/ASR/epoch_00080.pth"
+PLBERT_dir: 'Utils/PLBERT/'
+
+data_params:
+  train_data: "Data/train_harness.txt"
+  val_data: "Data/val_harness.txt"
+  root_path: ""
+  OOD_data: "Data/OOD_harness.txt"
+  min_length: 50 # sample until texts with this size are obtained for OOD texts
+
+preprocess_params:
+  sr: 24000
+  spect_params:
+    n_fft: 2048
+    win_length: 1200
+    hop_length: 300
+
+model_params:
+  multispeaker: true
+
+  dim_in: 64 
+  hidden_dim: 512
+  max_conv_dim: 512
+  n_layer: 3
+  n_mels: 80
+
+  n_token: 178 # number of phoneme tokens
+  max_dur: 50 # maximum duration of a single phoneme
+  style_dim: 128 # style vector size
+  
+  dropout: 0.2
+
+  # config for decoder
+  decoder: 
+      type: 'hifigan' # either hifigan or istftnet
+      resblock_kernel_sizes: [3,7,11]
+      upsample_rates :  [10,5,3,2]
+      upsample_initial_channel: 512
+      resblock_dilation_sizes: [[1,3,5], [1,3,5], [1,3,5]]
+      upsample_kernel_sizes: [20,10,6,4]
+      
+  # speech language model config
+  slm:
+      model: 'microsoft/wavlm-base-plus'
+      sr: 16000 # sampling rate of SLM
+      hidden: 768 # hidden size of SLM
+      nlayers: 13 # number of layers of SLM
+      initial_channel: 64 # initial channels of SLM discriminator head
+  
+  # style diffusion model config
+  diffusion:
+    embedding_mask_proba: 0.1
+    # transformer config
+    transformer:
+      num_layers: 3
+      num_heads: 8
+      head_features: 64
+      multiplier: 2
+
+    # diffusion distribution config
+    dist:
+      sigma_data: 0.2 # placeholder for estimate_sigma_data set to false
+      estimate_sigma_data: true # estimate sigma_data from the current batch if set to true
+      mean: -3.0
+      std: 1.0
+  
+loss_params:
+    lambda_mel: 5. # mel reconstruction loss
+    lambda_gen: 1. # generator loss
+    lambda_slm: 1. # slm feature matching loss
+    
+    lambda_mono: 1. # monotonic alignment loss (TMA)
+    lambda_s2s: 1. # sequence-to-sequence loss (TMA)
+
+    lambda_F0: 1. # F0 reconstruction loss
+    lambda_norm: 1. # norm reconstruction loss
+    lambda_dur: 1. # duration loss
+    lambda_ce: 20. # duration predictor probability output CE loss
+    lambda_sty: 1. # style reconstruction loss
+    lambda_diff: 1. # score matching loss
+    
+    diff_epoch: 10 # style diffusion starting epoch
+    joint_epoch: 30 # joint training starting epoch
+
+optimizer_params:
+  lr: 0.0001 # general learning rate
+  bert_lr: 0.00001 # learning rate for PLBERT
+  ft_lr: 0.0001 # learning rate for acoustic modules
+  
+slmadv_params:
+  min_len: 400 # minimum length of samples
+  max_len: 500 # maximum length of samples
+  batch_percentage: 0.5 # to prevent out of memory, only use half of the original batch size
+  iter: 10 # update the discriminator every this iterations of generator update
+  thresh: 5 # gradient norm above which the gradient is scaled
+  scale: 0.01 # gradient scaling factor for predictors from SLM discriminators
+  sig: 1.5 # sigma for differentiable duration modeling

tts/StyleTTS2Train/Models/LibriTTS/config_harness.yml

@@ -0,0 +1,111 @@
+log_dir: "Models/LJSpeech"
+save_freq: 5
+log_interval: 10
+device: "cuda"
+epochs: 50 # number of finetuning epoch (1 hour of data)
+batch_size: 1
+max_len: 400 # maximum number of frames
+pretrained_model: "Models/LibriTTS/epochs_2nd_00020.pth"
+second_stage_load_pretrained: true # set to true if the pre-trained model is for 2nd stage
+load_only_params: true # set to true if do not want to load epoch numbers and optimizer parameters
+
+F0_path: "Utils/JDC/bst.t7"
+ASR_config: "Utils/ASR/config.yml"
+ASR_path: "Utils/ASR/epoch_00080.pth"
+PLBERT_dir: 'Utils/PLBERT/'
+
+data_params:
+  train_data: "Data/train_harness.txt"
+  val_data: "Data/val_harness.txt"
+  root_path: ""
+  OOD_data: "Data/OOD_harness.txt"
+  min_length: 50 # sample until texts with this size are obtained for OOD texts
+
+preprocess_params:
+  sr: 24000
+  spect_params:
+    n_fft: 2048
+    win_length: 1200
+    hop_length: 300
+
+model_params:
+  multispeaker: true
+
+  dim_in: 64 
+  hidden_dim: 512
+  max_conv_dim: 512
+  n_layer: 3
+  n_mels: 80
+
+  n_token: 178 # number of phoneme tokens
+  max_dur: 50 # maximum duration of a single phoneme
+  style_dim: 128 # style vector size
+  
+  dropout: 0.2
+
+  # config for decoder
+  decoder: 
+      type: 'hifigan' # either hifigan or istftnet
+      resblock_kernel_sizes: [3,7,11]
+      upsample_rates :  [10,5,3,2]
+      upsample_initial_channel: 512
+      resblock_dilation_sizes: [[1,3,5], [1,3,5], [1,3,5]]
+      upsample_kernel_sizes: [20,10,6,4]
+      
+  # speech language model config
+  slm:
+      model: 'microsoft/wavlm-base-plus'
+      sr: 16000 # sampling rate of SLM
+      hidden: 768 # hidden size of SLM
+      nlayers: 13 # number of layers of SLM
+      initial_channel: 64 # initial channels of SLM discriminator head
+  
+  # style diffusion model config
+  diffusion:
+    embedding_mask_proba: 0.1
+    # transformer config
+    transformer:
+      num_layers: 3
+      num_heads: 8
+      head_features: 64
+      multiplier: 2
+
+    # diffusion distribution config
+    dist:
+      sigma_data: 0.2 # placeholder for estimate_sigma_data set to false
+      estimate_sigma_data: true # estimate sigma_data from the current batch if set to true
+      mean: -3.0
+      std: 1.0
+  
+loss_params:
+    lambda_mel: 5. # mel reconstruction loss
+    lambda_gen: 1. # generator loss
+    lambda_slm: 1. # slm feature matching loss
+    
+    lambda_mono: 1. # monotonic alignment loss (TMA)
+    lambda_s2s: 1. # sequence-to-sequence loss (TMA)
+
+    lambda_F0: 1. # F0 reconstruction loss
+    lambda_norm: 1. # norm reconstruction loss
+    lambda_dur: 1. # duration loss
+    lambda_ce: 20. # duration predictor probability output CE loss
+    lambda_sty: 1. # style reconstruction loss
+    lambda_diff: 1. # score matching loss
+    
+    diff_epoch: 10 # style diffusion starting epoch
+    joint_epoch: 30 # joint training starting epoch
+
+optimizer_params:
+  lr: 0.0001 # general learning rate
+  bert_lr: 0.00001 # learning rate for PLBERT
+  ft_lr: 0.0001 # learning rate for acoustic modules
+  
+slmadv_params:
+  min_len: 400 # minimum length of samples
+  max_len: 500 # maximum length of samples
+  batch_percentage: 0.5 # to prevent out of memory, only use half of the original batch size
+  iter: 10 # update the discriminator every this iterations of generator update
+  thresh: 5 # gradient norm above which the gradient is scaled
+  scale: 0.01 # gradient scaling factor for predictors from SLM discriminators
+  sig: 1.5 # sigma for differentiable duration modeling
+  

tts/StyleTTS2TrainOG/train_finetune.py

@@ -46,6 +46,8 @@ handler = StreamHandler()
 handler.setLevel(logging.DEBUG)
 logger.addHandler(handler)
 
+from test_harness import TestHarness
+harness = TestHarness()
 
 @click.command()
 @click.option('-p', '--config_path', default='Configs/config_ft.yml', type=str)
@@ -559,6 +561,13 @@ def main(config_path):
                 writer.add_scalar('train/gen_loss_slm', loss_gen_lm, iters)
                 
                 running_loss = 0
+                step_total = len(train_list)//batch_size
+                train_list_len = len(train_list)
+                step = i+1
+                
+                harness.log_values(loss_gen_all, d_loss, loss_ce, loss_dur, loss_lm, loss_norm_rec, 
+                   loss_F0_rec, loss_sty, loss_diff, d_loss_slm, loss_gen_lm, iters, epoch, step, 
+                   log_interval, running_loss,train_list_len,batch_size)
                 
                 print('Time elasped:', time.time()-start_time)
             
@@ -680,6 +689,7 @@ def main(config_path):
         writer.add_scalar('eval/dur_loss', loss_test / iters_test, epoch + 1)
         writer.add_scalar('eval/F0_loss', loss_f / iters_test, epoch + 1)
         
+        harness.log_eval(mel_loss=float(loss_test / iters_test),dur_loss=float(loss_test / iters_test),F0_loss=float(loss_f / iters_test),epoch=epoch)
         
         if (epoch + 1) % save_freq == 0 :
             if (loss_test / iters_test) < best_loss:
@@ -694,7 +704,8 @@ def main(config_path):
             }
             save_path = osp.join(log_dir, 'epoch_2nd_%05d.pth' % epoch)
             torch.save(state, save_path)
-
+            
+            harness.test(save_path)
             # if estimate sigma, save the estimated simga
             if model_params.diffusion.dist.estimate_sigma_data:
                 config['model_params']['diffusion']['dist']['sigma_data'] = float(np.mean(running_std))