FastPitch_Data_Preparation for single speaker requires to provide speaker id and speaker map path

[plamer777]

Hi there! I've faced an issue while starting FastPitch training using provided manual. I've successfully trained HiFiGAN but can't start training for FastPitch. It always asks to provide speaker id and path to mapper. After some manipulations I still have had an error: ValueError: You should add additional data types as conditions (e.g. speaker id or reference audio)
[rank0]: and define speaker_encoder in your config.
And finely I've got an error kind of the class (I've forgot the name) requires one positional argument but two was givien.
Are any ideas how to start the script for dataset with single speaker? Maybe there fastpitch.yaml with necessary settings or arguments I have to provide while starting the script?
Here're my args I use to start training:

` if torch.cuda.is_available():
accelerator = "gpu"
batch_size = 32
else:
accelerator = "cpu"
batch_size = 4

args = [
    f"--config-path={fastpitch_config_dir}",
    f"--config-name={fastpitch_config_filename}",
    f"n_speakers={num_speakers}",
    f"batch_size={batch_size}",
    f"max_epochs={epochs}",
    f"weighted_sampling_steps_per_epoch={steps_per_epoch}",
    f"phoneme_dict_path={phoneme_dict_path}",
    f"heteronyms_path={heteronyms_path}",
    f"feature_stats_path={stats_path}",
    f"log_dir={fastpitch_log_dir}",
    f"vocoder_type={vocoder_type}",
    f"vocoder_checkpoint_path=\'{vocoder_checkpoint_path}\'",
    f"trainer.accelerator={accelerator}",
    f"exp_manager.exp_dir={exp_dir}",
    f"+exp_manager.version={run_id}",
    f"+train_ds_meta.{dataset_name}.manifest_path={train_manifest_filepath}",
    f"+train_ds_meta.{dataset_name}.audio_dir={audio_dir}",
    f"+train_ds_meta.{dataset_name}.feature_dir={feature_dir}",
    f"+val_ds_meta.{dataset_name}.manifest_path={dev_manifest_filepath}",
    f"+val_ds_meta.{dataset_name}.audio_dir={audio_dir}",
    f"+val_ds_meta.{dataset_name}.feature_dir={feature_dir}",
    f"+log_ds_meta.{dataset_name}.manifest_path={dev_manifest_filepath}",
    f"+log_ds_meta.{dataset_name}.audio_dir={audio_dir}",
    f"+log_ds_meta.{dataset_name}.feature_dir={feature_dir}"
]
if num_speakers > 1:
    if not speaker_path:
        raise ValueError(
            "speaker_path must be provided if num_speakers > 1"
        )
    args.append(f"speaker_path={speaker_path}")

run_script(fastpitch_training_script, args)`

Very appreciate for any help!