[Bug] AttributeError: 'StochasticDurationPredictor' object has no attribute 'cond'

[OmiWakode]

🐛 Description

AttributeError: 'StochasticDurationPredictor' object has no attribute 'cond' I have got this error while inferencing from the vits model, for zero shot learning. I have the following function for synthesis and have replaced d_vector with a speaker file whom I want to learn by zero shot cloning. (reference_emb is a d_vector generated from speaker encoder) I have used the script from here for inferencing.

wav, alignment, _, _ = synthesis(
                    model,
                    text,
                    C,
                    "cuda" in str(next(model.parameters()).device),
                    speaker_id = None,
                    d_vector = reference_emb,
                    style_wav=None,
                    language_id=None,
                    #enable_eos_bos_chars=C.enable_eos_bos_chars,
                    use_griffin_lim= False,
                    do_trim_silence=False,
                ).values()

Traceback of the error

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-25-9ebb3c30e311> in <module>()
     13                     #enable_eos_bos_chars=C.enable_eos_bos_chars,
     14                     use_griffin_lim= False,
---> 15                     do_trim_silence=False,
     16                 ).values()
     17 print("Generated Audio")

5 frames
/content/drive/MyDrive/coqui-cloned-repo/TTS/TTS/tts/utils/synthesis.py in synthesis(model, text, CONFIG, use_cuda, speaker_id, style_wav, use_griffin_lim, do_trim_silence, d_vector, language_id)
    184     text_inputs = text_inputs.unsqueeze(0)
    185     # synthesize voice
--> 186     outputs = run_model_torch(model, text_inputs, speaker_id, style_mel, d_vector=d_vector, language_id=language_id)
    187     model_outputs = outputs["model_outputs"]
    188     model_outputs = model_outputs[0].data.cpu().numpy()

/content/drive/MyDrive/coqui-cloned-repo/TTS/TTS/tts/utils/synthesis.py in run_model_torch(model, inputs, speaker_id, style_mel, d_vector, language_id)
     54             "d_vectors": d_vector,
     55             "style_mel": style_mel,
---> 56             "language_ids": language_id,
     57         },
     58     )

/content/drive/MyDrive/coqui-cloned-repo/TTS/TTS/tts/models/vits.py in inference(self, x, aux_input)
    972                 reverse=True,
    973                 noise_scale=self.inference_noise_scale_dp,
--> 974                 lang_emb=lang_emb,
    975             )
    976         else:

/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
   1049         if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1050                 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1051             return forward_call(*input, **kwargs)
   1052         # Do not call functions when jit is used
   1053         full_backward_hooks, non_full_backward_hooks = [], []

/content/drive/MyDrive/coqui-cloned-repo/TTS/TTS/tts/layers/vits/stochastic_duration_predictor.py in forward(self, x, x_mask, dr, g, lang_emb, reverse, noise_scale)
    231         x = self.pre(x)
    232         if g is not None:
--> 233             x = x + self.cond(g)
    234 
    235         if lang_emb is not None:

/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py in __getattr__(self, name)
   1129                 return modules[name]
   1130         raise AttributeError("'{}' object has no attribute '{}'".format(
-> 1131             type(self).__name__, name))
   1132 
   1133     def __setattr__(self, name: str, value: Union[Tensor, 'Module']) -> None:

AttributeError: 'StochasticDurationPredictor' object has no attribute 'cond'

Additional Information:

I have trained my Vits Model with a separate d_vector_file, and I want to make my custom YourTTS on my own data.

• 🐸TTS Version : v0.6.1

• GPU models and configuration: Colab (Free)

• Installed everything as mentioned in the above colab link

[Edresson]

Hi @OmiWakode ,

This notebook is old and compatible just with the Edresson/Coqui-TTS@multilingual-torchaudio-SE branch.

Currently, the TTS CLI works with the released YourTTS model. After install the latest version of Coqui TTS, You can do zero-shot multi-speaker TTS using the following command:

tts --model_name "tts_models/multilingual/multi-dataset/your_tts" --text "This is an example" --speaker_wav ../ref.wav --language_idx "en" --out_path ../output_test.wav

Just update the text, speaker_wav (a waveform with the voice of the target speaker), and out_path :).

[OmiWakode]

Can I use this same script for zero shot cloning with my custom vits model?

[OmiWakode]

I want to perform zero shot cloning with my custom vits model, or maybe try to follow YourTTS inference to my custom vits.

[Edresson]

Can I use this same script for zero shot cloning with my custom vits model?

Yeah you can , but you need to provide remove the "--model_name" and add other params like: --model_path, --config_path, --speakers_file_path, and --language_ids_file_path (just if it is a multilingual model)

[OmiWakode]

Hey, Tried that, again got the same error: Here is my code

!tts --config_path /config.json
 --model_path best_model.pth 
--speakers_file_path speakers.json 
--text "This is an example" 
--speaker_wav /content/drive/MyDrive/zero-shot-coqui-ai/videoplayback_1hr.wav 
--out_path ../output_test.wav

Traceback (most recent call last):
  File "/usr/local/bin/tts", line 8, in <module>
    sys.exit(main())
  File "/usr/local/lib/python3.7/dist-packages/TTS/bin/synthesize.py", line 287, in main
    wav = synthesizer.tts(args.text, args.speaker_idx, args.language_idx, args.speaker_wav)
  File "/usr/local/lib/python3.7/dist-packages/TTS/utils/synthesizer.py", line 260, in tts
    d_vector=speaker_embedding,
  File "/usr/local/lib/python3.7/dist-packages/TTS/tts/utils/synthesis.py", line 184, in synthesis
    outputs = run_model_torch(model, text_inputs, speaker_id, style_mel, d_vector=d_vector, language_id=language_id)
  File "/usr/local/lib/python3.7/dist-packages/TTS/tts/utils/synthesis.py", line 56, in run_model_torch
    "language_ids": language_id,
  File "/usr/local/lib/python3.7/dist-packages/TTS/tts/models/vits.py", line 974, in inference
    lang_emb=lang_emb,
  File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/usr/local/lib/python3.7/dist-packages/TTS/tts/layers/vits/stochastic_duration_predictor.py", line 233, in forward
    x = x + self.cond(g)
  File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1178, in __getattr__
    type(self).__name__, name))
AttributeError: 'StochasticDurationPredictor' object has no attribute 'cond'

[Edresson]

Could you share your config.json?

[OmiWakode]

Sure

{
    "output_path": "/home/ubuntu/train_pipeline/ReChannelisedData/outputs",
    "logger_uri": null,
    "run_name": "last_vits",
    "project_name": null,
    "run_description": "\ud83d\udc38Coqui trainer run.",
    "print_step": 25,
    "plot_step": 100,
    "model_param_stats": false,
    "wandb_entity": null,
    "dashboard_logger": "tensorboard",
    "log_model_step": 10000,
    "save_step": 10000,
    "save_n_checkpoints": 5,
    "save_checkpoints": true,
    "save_all_best": false,
    "save_best_after": 10000,
    "target_loss": null,
    "print_eval": false,
    "test_delay_epochs": -1,
    "run_eval": true,
    "distributed_backend": "nccl",
    "distributed_url": "tcp://localhost:54321",
    "mixed_precision": true,
    "epochs": 10,
    "batch_size": 32,
    "eval_batch_size": 16,
    "grad_clip": [
        1000,
        1000
    ],
    "scheduler_after_epoch": true,
    "lr": 0.001,
    "optimizer": "AdamW",
    "optimizer_params": {
        "betas": [
            0.8,
            0.99
        ],
        "eps": 1e-09,
        "weight_decay": 0.01
    },
    "lr_scheduler": "",
    "lr_scheduler_params": {},
    "use_grad_scaler": false,
    "cudnn_enable": true,
    "cudnn_benchmark": true,
    "torch_seed": 54321,
    "model": "vits",
    "num_loader_workers": 4,
    "num_eval_loader_workers": 4,
    "use_noise_augment": false,
    "audio": {
        "fft_size": 1024,
        "win_length": 1024,
        "hop_length": 256,
        "frame_shift_ms": null,
        "frame_length_ms": null,
        "stft_pad_mode": "reflect",
        "sample_rate": 22050,
        "resample": true,
        "preemphasis": 0.0,
        "ref_level_db": 20,
        "do_sound_norm": false,
        "log_func": "np.log",
        "do_trim_silence": true,
        "trim_db": 23.0,
        "do_rms_norm": false,
        "db_level": null,
        "power": 1.5,
        "griffin_lim_iters": 60,
        "num_mels": 80,
        "mel_fmin": 0,
        "mel_fmax": null,
        "spec_gain": 1.0,
        "do_amp_to_db_linear": false,
        "do_amp_to_db_mel": true,
        "pitch_fmax": 640.0,
        "pitch_fmin": 0.0,
        "signal_norm": false,
        "min_level_db": -100,
        "symmetric_norm": true,
        "max_norm": 4.0,
        "clip_norm": true,
        "stats_path": null
    },
    "use_phonemes": false,
    "phonemizer": null,
    "phoneme_language": "en-us",
    "compute_input_seq_cache": true,
    "text_cleaner": "english_cleaners",
    "enable_eos_bos_chars": false,
    "test_sentences_file": "",
    "phoneme_cache_path": null,
    "characters": {
        "characters_class": "TTS.tts.utils.text.characters.Graphemes",
        "vocab_dict": null,
        "pad": "<PAD>",
        "eos": "<EOS>",
        "bos": "<BOS>",
        "blank": "<BLNK>",
        "characters": "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz",
        "punctuations": "!'(),-.:;? ",
        "phonemes": null,
        "is_unique": false,
        "is_sorted": true
    },
    "add_blank": true,
    "batch_group_size": 0,
    "loss_masking": null,
    "sort_by_audio_len": false,
    "min_audio_len": 1,
    "max_audio_len": Infinity,
    "min_text_len": 1,
    "max_text_len": 325,
    "compute_f0": false,
    "compute_linear_spec": true,
    "precompute_num_workers": 0,
    "start_by_longest": false,
    "datasets": [
        {
            "name": "indian_english_speakers",
            "path": "/home/ubuntu/train_pipeline/ReChannelisedData/",
            "meta_file_train": "metadata.csv",
            "ignored_speakers": null,
            "language": "en-us",
            "meta_file_val": "",
            "meta_file_attn_mask": ""
        }
    ],
    "test_sentences": [
        [
            "It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent."
        ],
        [
            "Be a voice, not an echo."
        ],
        [
            "I'm sorry Dave. I'm afraid I can't do that."
        ],
        [
            "This cake is great. It's so delicious and moist."
        ],
        [
            "Prior to November 22, 1963."
        ]
    ],
    "eval_split_max_size": null,
    "eval_split_size": 0.01,
    "use_speaker_weighted_sampler": false,
    "speaker_weighted_sampler_alpha": 1.0,
    "use_language_weighted_sampler": false,
    "language_weighted_sampler_alpha": 1.0,
    "model_args": {
        "num_chars": 67,
        "out_channels": 513,
        "spec_segment_size": 32,
        "hidden_channels": 192,
        "hidden_channels_ffn_text_encoder": 768,
        "num_heads_text_encoder": 2,
        "num_layers_text_encoder": 6,
        "kernel_size_text_encoder": 3,
        "dropout_p_text_encoder": 0.1,
        "dropout_p_duration_predictor": 0.5,
        "kernel_size_posterior_encoder": 5,
        "dilation_rate_posterior_encoder": 1,
        "num_layers_posterior_encoder": 16,
        "kernel_size_flow": 5,
        "dilation_rate_flow": 1,
        "num_layers_flow": 4,
        "resblock_type_decoder": "1",
        "resblock_kernel_sizes_decoder": [
            3,
            7,
            11
        ],
        "resblock_dilation_sizes_decoder": [
            [
                1,
                3,
                5
            ],
            [
                1,
                3,
                5
            ],
            [
                1,
                3,
                5
            ]
        ],
        "upsample_rates_decoder": [
            8,
            8,
            2,
            2
        ],
        "upsample_initial_channel_decoder": 512,
        "upsample_kernel_sizes_decoder": [
            16,
            16,
            4,
            4
        ],
        "use_sdp": true,
        "noise_scale": 1.0,
        "inference_noise_scale": 0.667,
        "length_scale": 1,
        "noise_scale_dp": 1.0,
        "inference_noise_scale_dp": 1.0,
        "max_inference_len": null,
        "init_discriminator": true,
        "use_spectral_norm_disriminator": false,
        "use_speaker_embedding": false,
        "num_speakers": 10,
        "speakers_file": "/home/ubuntu/train_pipeline/ReChannelisedData/outputs/last_vits-March-29-2022_03+33PM-0000000/speakers.json",
        "d_vector_file": "/content/drive/MyDrive/Custom_coqui/MyTTSDataset/sample_indian_embed_try.json",
        "speaker_embedding_channels": 256,
        "use_d_vector_file": true,
        "d_vector_dim": 0,
        "detach_dp_input": true,
        "use_language_embedding": false,
        "embedded_language_dim": 4,
        "num_languages": 0,
        "language_ids_file": null,
        "use_speaker_encoder_as_loss": false,
        "speaker_encoder_config_path": "/content/drive/MyDrive/Custom_coqui/ASP-oficial-model-20220314T094714Z-001/H-ASP-oficial-model/config.json",
        "speaker_encoder_model_path": "/content/drive/MyDrive/Custom_coqui/ASP-oficial-model-20220314T094714Z-001/H-ASP-oficial-model/converted_checkpoint.pth.tar",
        "condition_dp_on_speaker": true,
        "freeze_encoder": false,
        "freeze_DP": false,
        "freeze_PE": false,
        "freeze_flow_decoder": false,
        "freeze_waveform_decoder": false
    },
    "lr_gen": 0.0002,
    "lr_disc": 0.0002,
    "lr_scheduler_gen": "ExponentialLR",
    "lr_scheduler_gen_params": {
        "gamma": 0.999875,
        "last_epoch": -1
    },
    "lr_scheduler_disc": "ExponentialLR",
    "lr_scheduler_disc_params": {
        "gamma": 0.999875,
        "last_epoch": -1
    },
    "kl_loss_alpha": 1.0,
    "disc_loss_alpha": 1.0,
    "gen_loss_alpha": 1.0,
    "feat_loss_alpha": 1.0,
    "mel_loss_alpha": 45.0,
    "dur_loss_alpha": 1.0,
    "speaker_encoder_loss_alpha": 1.0,
    "return_wav": true,
    "r": 1,
    "num_speakers": 0,
    "use_speaker_embedding": false,
    "speakers_file": "/home/ubuntu/train_pipeline/ReChannelisedData/outputs/last_vits-March-29-2022_03+33PM-0000000/speakers.json",
    "speaker_embedding_channels": 256,
    "language_ids_file": null,
    "use_language_embedding": false,
    "use_d_vector_file": true,
    "d_vector_file": "/content/drive/MyDrive/Custom_coqui/MyTTSDataset/sample_indian_embed_try.json",
    "d_vector_dim": 0
}

[Edresson]

Hi @OmiWakode You need to define d_vector_dim in model_args and in the general config (at the end of the config.json file). Currently, in your config.json "d_vector_dim" is equal to zero. In this way, the model will create the layers necessary to condition the speaker embedding. Please set "d_vector_dim" to the external speaker embedding dim in all apparitions in the config file.

[shigabeev]

The issue is still an issue by the way.

[shigabeev]

Hey! I'm trying to run models with speaker consistency loss and the inference doesn't run:

import torch
from TTS.tts.utils.synthesis import synthesis

model_root = path/to/model
C = load_config(model_root + '/config.json')
state_dict = torch.load(model_root + '/best_model.pth')
model = setup_model(config=C)
model.load_state_dict(state_dict['model'])

res = synthesis(model, text="This is an example of the voice produced by multi speaker model.", 
          CONFIG=C, 
          use_cuda=False,
          d_vector=torch.randn(512),
          )

It returns the same error as in here.

AttributeError: 'StochasticDurationPredictor' object has no attribute 'cond'

Fun fact: defining d_vector_dim doesn't help anymore.

When I change config to:

C.model_args['use_d_vector_file'] = True
C.model_args['d_vector_file'] = TTS_SPEAKERS

it allows to partially load weights of model with strict=false: model.load_state_dict(model_weights, strict=False) But the inference is of understandably terrible quality since some layers didn't load and contain pure noise.

Another option is to change config with:

C.model_args['d_vector_file'] = None
C.model_args['speaker_encoder_model_path'] = None
C.model_args['speaker_encoder_config_path'] = None
C.model_args['use_speaker_encoder_as_loss'] = False
C.model_args['use_d_vector_file'] = False
C.use_speaker_embedding = True

This way, weights will load without raising error, but since there's no trace of speaker encoder left in model, it would raise the same error as in OPs case:

AttributeError: 'StochasticDurationPredictor' object has no attribute 'cond'

Steps to reproduce:

1. launch the config for a few steps, i.e. 2, it would generate weights for the model.
2. Run inference through CLI/Python.

CoquiTTS version: 0.8.0 My model config:

{
    "output_path": "/home/frappuccino/recipes/out/",
    "logger_uri": null,
    "run_name": "vits_spk_consist",
    "project_name": null,
    "run_description": "\ud83d\udc38Coqui trainer run.",
    "print_step": 25,
    "plot_step": 100,
    "model_param_stats": false,
    "wandb_entity": null,
    "dashboard_logger": "tensorboard",
    "log_model_step": 10000,
    "save_step": 10000,
    "save_n_checkpoints": 5,
    "save_checkpoints": true,
    "save_all_best": false,
    "save_best_after": 10000,
    "target_loss": null,
    "print_eval": false,
    "test_delay_epochs": -1,
    "run_eval": true,
    "run_eval_steps": null,
    "distributed_backend": "nccl",
    "distributed_url": "tcp://localhost:54321",
    "mixed_precision": false,
    "epochs": 1000,
    "batch_size": 32,
    "eval_batch_size": 32,
    "grad_clip": [
        5.0,
        5.0
    ],
    "scheduler_after_epoch": true,
    "lr": 0.001,
    "optimizer": "AdamW",
    "optimizer_params": {
        "betas": [
            0.8,
            0.99
        ],
        "eps": 1e-09,
        "weight_decay": 0.01
    },
    "lr_scheduler": "",
    "lr_scheduler_params": {},
    "use_grad_scaler": false,
    "cudnn_enable": true,
    "cudnn_deterministic": false,
    "cudnn_benchmark": false,
    "training_seed": 54321,
    "model": "vits",
    "num_loader_workers": 12,
    "num_eval_loader_workers": 1,
    "use_noise_augment": false,
    "audio": {
        "fft_size": 1024,
        "sample_rate": 16000,
        "win_length": 1024,
        "hop_length": 256,
        "num_mels": 80,
        "mel_fmin": 0,
        "mel_fmax": null
    },
    "use_phonemes": true,
    "phonemizer": "espeak",
    "phoneme_language": "en",
    "compute_input_seq_cache": true,
    "text_cleaner": "english_cleaners",
    "enable_eos_bos_chars": false,
    "test_sentences_file": "",
    "phoneme_cache_path": "/home/frappuccino/recipes/out/phoneme_cache",
    "characters": {
        "characters_class": "TTS.tts.utils.text.characters.IPAPhonemes",
        "vocab_dict": null,
        "pad": "#",
        "eos": "$",
        "bos": "^",
        "blank": "_",
        "characters": "iy\u0268\u0289\u026fu\u026a\u028f\u028ae\u00f8\u0258\u0259\u0275\u0264o\u025b\u0153\u025c\u025e\u028c\u0254\u00e6\u0250a\u0276\u0251\u0252\u1d7b\u0298\u0253\u01c0\u0257\u01c3\u0284\u01c2\u0260\u01c1\u029bpbtd\u0288\u0256c\u025fk\u0261q\u0262\u0294\u0274\u014b\u0272\u0273n\u0271m\u0299r\u0280\u2c71\u027e\u027d\u0278\u03b2fv\u03b8\u00f0sz\u0283\u0292\u0282\u0290\u00e7\u029dx\u0263\u03c7\u0281\u0127\u0295h\u0266\u026c\u026e\u028b\u0279\u027bj\u0270l\u026d\u028e\u029f\u02c8\u02cc\u02d0\u02d1\u028dw\u0265\u029c\u02a2\u02a1\u0255\u0291\u027a\u0267\u02b2\u025a\u02de\u026b",
        "punctuations": "!'(),-.:;? ",
        "phonemes": null,
        "is_unique": true,
        "is_sorted": true
    },
    "add_blank": true,
    "batch_group_size": 0,
    "loss_masking": null,
    "min_audio_len": 1,
    "max_audio_len": Infinity,
    "min_text_len": 1,
    "max_text_len": 500,
    "compute_f0": false,
    "compute_linear_spec": true,
    "precompute_num_workers": 0,
    "start_by_longest": false,
    "datasets": [
        {
            "name": "libri_tts",
            "path": "/home/frappuccino/data/librispeech16k/train-clean-360-16k/",
            "meta_file_train": "",
            "ignored_speakers": null,
            "language": "en",
            "meta_file_val": "",
            "meta_file_attn_mask": ""
        }
    ],
    "test_sentences": [
        [
            "It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent."
        ],
        [
            "Be a voice, not an echo."
        ],
        [
            "I'm sorry Dave. I'm afraid I can't do that."
        ],
        [
            "This cake is great. It's so delicious and moist."
        ],
        [
            "Prior to November 22, 1963."
        ]
    ],
    "eval_split_max_size": null,
    "eval_split_size": 0.01,
    "use_speaker_weighted_sampler": false,
    "speaker_weighted_sampler_alpha": 1.0,
    "use_language_weighted_sampler": false,
    "language_weighted_sampler_alpha": 1.0,
    "use_length_weighted_sampler": false,
    "length_weighted_sampler_alpha": 1.0,
    "model_args": {
        "num_chars": 131,
        "out_channels": 513,
        "spec_segment_size": 62,
        "hidden_channels": 192,
        "hidden_channels_ffn_text_encoder": 768,
        "num_heads_text_encoder": 2,
        "num_layers_text_encoder": 10,
        "kernel_size_text_encoder": 3,
        "dropout_p_text_encoder": 0.1,
        "dropout_p_duration_predictor": 0.5,
        "kernel_size_posterior_encoder": 5,
        "dilation_rate_posterior_encoder": 1,
        "num_layers_posterior_encoder": 16,
        "kernel_size_flow": 5,
        "dilation_rate_flow": 1,
        "num_layers_flow": 4,
        "resblock_type_decoder": "2",
        "resblock_kernel_sizes_decoder": [
            3,
            7,
            11
        ],
        "resblock_dilation_sizes_decoder": [
            [
                1,
                3,
                5
            ],
            [
                1,
                3,
                5
            ],
            [
                1,
                3,
                5
            ]
        ],
        "upsample_rates_decoder": [
            8,
            8,
            2,
            2
        ],
        "upsample_initial_channel_decoder": 512,
        "upsample_kernel_sizes_decoder": [
            16,
            16,
            4,
            4
        ],
        "periods_multi_period_discriminator": [
            2,
            3,
            5,
            7,
            11
        ],
        "use_sdp": true,
        "noise_scale": 1.0,
        "inference_noise_scale": 0.3,
        "length_scale": 1.5,
        "noise_scale_dp": 0.6,
        "inference_noise_scale_dp": 0.3,
        "max_inference_len": null,
        "init_discriminator": true,
        "use_spectral_norm_disriminator": false,
        "use_speaker_embedding": false,
        "num_speakers": 902,
        "speakers_file": "out/vits_spk_consist-October-20-2022_07+43AM-9017f31/speakers.pth",
        "d_vector_file": null,
        "speaker_embedding_channels": 512,
        "use_d_vector_file": false,
        "d_vector_dim": 512,
        "detach_dp_input": true,
        "use_language_embedding": false,
        "embedded_language_dim": 4,
        "num_languages": 0,
        "language_ids_file": null,
        "use_speaker_encoder_as_loss": true,
        "speaker_encoder_config_path": "config_se.json",
        "speaker_encoder_model_path": "SE_checkpoint.pth.tar",
        "condition_dp_on_speaker": true,
        "freeze_encoder": false,
        "freeze_DP": false,
        "freeze_PE": false,
        "freeze_flow_decoder": false,
        "freeze_waveform_decoder": false,
        "encoder_sample_rate": null,
        "interpolate_z": true,
        "reinit_DP": false,
        "reinit_text_encoder": false
    },
    "lr_gen": 0.0002,
    "lr_disc": 0.0002,
    "lr_scheduler_gen": "ExponentialLR",
    "lr_scheduler_gen_params": {
        "gamma": 0.999875,
        "last_epoch": -1
    },
    "lr_scheduler_disc": "ExponentialLR",
    "lr_scheduler_disc_params": {
        "gamma": 0.999875,
        "last_epoch": -1
    },
    "kl_loss_alpha": 1.0,
    "disc_loss_alpha": 1.0,
    "gen_loss_alpha": 1.0,
    "feat_loss_alpha": 1.0,
    "mel_loss_alpha": 45.0,
    "dur_loss_alpha": 1.0,
    "speaker_encoder_loss_alpha": 9.0,
    "return_wav": true,
    "use_weighted_sampler": false,
    "weighted_sampler_attrs": {},
    "weighted_sampler_multipliers": {},
    "r": 1,
    "num_speakers": 902,
    "use_speaker_embedding": false,
    "speakers_file": "out/vits_spk_consist-October-20-2022_07+43AM-9017f31/speakers.pth",
    "speaker_embedding_channels": 512,
    "language_ids_file": null,
    "use_language_embedding": false,
    "use_d_vector_file": false,
    "d_vector_file": null,
    "d_vector_dim": 512
}