o jr@s`ddlmZmZmZddlmZmZddlmZm Z ddl m Z m Z m Z eGddde ZdS))asdict dataclassfield)DictList)Coqpitcheck_argument)BaseAudioConfigBaseDatasetConfigBaseTrainingConfigc@seZdZUdZeedZeed<dZe ed<dZ e ed<dZ e ed <e d Ze ed <dZe ed <e d Ze ed <dZe ed<dZe ed<dZe ed<dZe ed<dZe ed<dZe ed<dZe ed<dZe ed<edddZeeed<dZeed<dZeed<dZ eed<edddZ!eed<ed ddZ"eeed!<dZ#e ed"<d#Z$e ed$<dZ%e ed%<d&Z&e ed'<dZ'e ed(<d&Z(e ed)<dZ)e ed*<d&Z*e ed+<dS), BaseVCConfiga6Shared parameters among all the tts models. Args: audio (BaseAudioConfig): Audio processor config object instance. batch_group_size (int): Size of the batch groups used for bucketing. By default, the dataloader orders samples by the sequence length for a more efficient and stable training. If `batch_group_size > 1` then it performs bucketing to prevent using the same batches for each epoch. loss_masking (bool): enable / disable masking loss values against padded segments of samples in a batch. min_text_len (int): Minimum length of input text to be used. All shorter samples will be ignored. Defaults to 0. max_text_len (int): Maximum length of input text to be used. All longer samples will be ignored. Defaults to float("inf"). min_audio_len (int): Minimum length of input audio to be used. All shorter samples will be ignored. Defaults to 0. max_audio_len (int): Maximum length of input audio to be used. All longer samples will be ignored. The maximum length in the dataset defines the VRAM used in the training. Hence, pay attention to this value if you encounter an OOM error in training. Defaults to float("inf"). compute_f0 (int): (Not in use yet). compute_energy (int): (Not in use yet). compute_linear_spec (bool): If True data loader computes and returns linear spectrograms alongside the other data. precompute_num_workers (int): Number of workers to precompute features. Defaults to 0. use_noise_augment (bool): Augment the input audio with random noise. start_by_longest (bool): If True, the data loader will start loading the longest batch first. It is useful for checking OOM issues. Defaults to False. shuffle (bool): If True, the data loader will shuffle the dataset when there is not sampler defined. Defaults to True. drop_last (bool): If True, the data loader will drop the last batch if it is not complete. It helps to prevent issues that emerge from the partial batch statistics. Defaults to True. add_blank (bool): Add blank characters between each other two characters. It improves performance for some models at expense of slower run-time due to the longer input sequence. datasets (List[BaseDatasetConfig]): List of datasets used for training. If multiple datasets are provided, they are merged and used together for training. optimizer (str): Optimizer used for the training. Set one from `torch.optim.Optimizer` or `TTS.utils.training`. Defaults to ``. optimizer_params (dict): Optimizer kwargs. Defaults to `{"betas": [0.8, 0.99], "weight_decay": 0.0}` lr_scheduler (str): Learning rate scheduler for the training. Use one from `torch.optim.Scheduler` schedulers or `TTS.utils.training`. Defaults to ``. lr_scheduler_params (dict): Parameters for the generator learning rate scheduler. Defaults to `{"warmup": 4000}`. test_sentences (List[str]): List of sentences to be used at testing. Defaults to '[]' eval_split_max_size (int): Number maximum of samples to be used for evaluation in proportion split. Defaults to None (Disabled). eval_split_size (float): If between 0.0 and 1.0 represents the proportion of the dataset to include in the evaluation set. If > 1, represents the absolute number of evaluation samples. Defaults to 0.01 (1%). use_speaker_weighted_sampler (bool): Enable / Disable the batch balancer by speaker. Defaults to ```False```. speaker_weighted_sampler_alpha (float): Number that control the influence of the speaker sampler weights. Defaults to ```1.0```. use_language_weighted_sampler (bool): Enable / Disable the batch balancer by language. Defaults to ```False```. language_weighted_sampler_alpha (float): Number that control the influence of the language sampler weights. Defaults to ```1.0```. use_length_weighted_sampler (bool): Enable / Disable the batch balancer by audio length. If enabled the dataset will be divided into 10 buckets considering the min and max audio of the dataset. The sampler weights will be computed forcing to have the same quantity of data for each bucket in each training batch. Defaults to ```False```. length_weighted_sampler_alpha (float): Number that control the influence of the length sampler weights. Defaults to ```1.0```. )default_factoryaudiorbatch_group_sizeN loss_masking min_audio_leninf max_audio_len min_text_len max_text_lenF compute_f0compute_energycompute_linear_specprecompute_num_workersuse_noise_augmentstart_by_longestshuffle drop_lastcCstgSN)r r r P/home/kuhnn/.local/lib/python3.10/site-packages/TTS/vc/configs/shared_configs.pyszBaseVCConfig.datasetsradam optimizeroptimizer_params lr_schedulercCsiSrr r r r r!r"lr_scheduler_paramscCsgSrr r r r r!r"r(test_sentenceseval_split_max_sizeg{Gz?eval_split_sizeuse_speaker_weighted_samplerg?speaker_weighted_sampler_alphause_language_weighted_samplerlanguage_weighted_sampler_alphause_length_weighted_samplerlength_weighted_sampler_alpha)+__name__ __module__ __qualname____doc__rr r__annotations__rintrboolrfloatrrrrrrrrrrrr#rr r%strr&dictr'r)r*r+r,r-r.r/r0r1r2r r r r!r s> l                      r N) dataclassesrrrtypingrrcoqpitrr TTS.configr r r r r r r r!s