Help me!!!!

[Mr47121836]

There is a problem with my program. The length of the dataloader is 0. after debugging, I found that my train datasets have data, but it can't be passed into the dataloader. Here is my program, please help me.

import os
import torch
from trainer import Trainer, TrainerArgs
from TTS.bin.compute_embeddings import compute_embeddings
from TTS.config.shared_configs import BaseDatasetConfig
from TTS.tts.configs.vits_config import VitsConfig
from TTS.tts.datasets import load_tts_samples
from TTS.tts.models.vits import CharactersConfig, Vits, VitsArgs, VitsAudioConfig
from TTS.utils.audio import AudioProcessor
from TTS.tts.datasets import my_data_formatter

# 使用当前脚本所在目录作为输出路径
OUT_PATH = os.path.dirname(os.path.abspath(__file__))

# 定义数据集配置
yourtts_config = BaseDatasetConfig(
    formatter="my_data_formatter",
    dataset_name="yourtts",
    meta_file_train="metadata.csv",
    path="/data/datasets/yourtts/2/",
    language="zh",
)
DATASETS_CONFIG_LIST = [yourtts_config]

# 提取说话人嵌入
SPEAKER_ENCODER_CHECKPOINT_PATH = "model_se.pth.tar"
SPEAKER_ENCODER_CONFIG_PATH = "config_se.json"
D_VECTOR_FILES = []

for dataset_conf in DATASETS_CONFIG_LIST:
    embeddings_file = os.path.join(dataset_conf.path, "speakers.pth")
    print(f">>> Computing the speaker embeddings for the {dataset_conf.dataset_name} dataset")
    compute_embeddings(
            SPEAKER_ENCODER_CHECKPOINT_PATH,
            SPEAKER_ENCODER_CONFIG_PATH,
            embeddings_file,
            old_speakers_file=None,
            config_dataset_path=None,
            formatter_name=dataset_conf.formatter,
            dataset_name=dataset_conf.dataset_name,
            dataset_path=dataset_conf.path,
            meta_file_train=dataset_conf.meta_file_train,
            meta_file_val=dataset_conf.meta_file_val,
            disable_cuda=False,
            no_eval=False,
        )
    D_VECTOR_FILES.append(embeddings_file)

# 音频配置
audio_config = VitsAudioConfig(
    sample_rate=22050,
    hop_length=256,
    win_length=1024,
    fft_size=1024,
    mel_fmin=0.0,
    mel_fmax=None,
    num_mels=80,
)

# YourTTS 的模型参数
model_args = VitsArgs(
    d_vector_file=D_VECTOR_FILES,
    use_d_vector_file=True,
    d_vector_dim=512,
    num_layers_text_encoder=10,
    speaker_encoder_model_path=SPEAKER_ENCODER_CHECKPOINT_PATH,
    speaker_encoder_config_path=SPEAKER_ENCODER_CONFIG_PATH,
    resblock_type_decoder="2",
    use_language_embedding=True,
    embedded_language_dim=4,
)

# 训练配置
config = VitsConfig(
    output_path=OUT_PATH,
    model_args=model_args,
    run_name="YourTTS-Chinese",
    project_name="YourTTS",
    run_description="YourTTS 中文数据集训练",
    dashboard_logger="tensorboard",
    logger_uri=None,
    audio=audio_config,
    batch_size=32,
    batch_group_size=48,
    eval_batch_size=32,
    num_loader_workers=4,
    eval_split_max_size=256,
    print_step=50,
    plot_step=100,
    log_model_step=1000,
    save_step=5000,
    save_n_checkpoints=2,
    save_checkpoints=True,
    target_loss="loss_1",
    print_eval=False,
    use_phonemes=True,
    phonemizer="espeak",
    phoneme_language="zh",
    compute_input_seq_cache=True,
    add_blank=True,
    text_cleaner="chinese_mandarin_cleaners",
    characters=CharactersConfig(
        characters_class="TTS.tts.models.vits.VitsCharacters",
        pad="_",
        eos="&",
        bos="*",
        blank=None,
        characters="ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\u00af\u00b7\u00df\u00e0\u00e1\u00e2\u00e3\u00e4\u00e6\u00e7\u00e8\u00e9\u00ea\u00eb\u00ec\u00ed\u00ee\u00ef\u00f1\u00f2\u00f3\u00f4\u00f5\u00f6\u00f9\u00fa\u00fb\u00fc\u00ff\u0101\u0105\u0107\u0113\u0119\u011b\u012b\u0131\u0142\u0144\u014d\u0151\u0153\u015b\u016b\u0171\u017a\u017c\u01ce\u01d0\u01d2\u01d4\u0430\u0431\u0432\u0433\u0434\u0435\u0436\u0437\u0438\u0439\u043a\u043b\u043c\u043d\u043e\u043f\u0440\u0441\u0442\u0443\u0444\u0445\u0446\u0447\u0448\u0449\u044a\u044b\u044c\u044d\u044e\u044f\u0451\u0454\u0456\u0457\u0491\u2013!'(),-.:;? ",
        punctuations="!'(),-.:;? ",
        phonemes="",
        is_unique=True,
        is_sorted=True,
    ),
    phoneme_cache_path=os.path.join(OUT_PATH, "phoneme_cache"),
    precompute_num_workers=12,
    start_by_longest=True,
    datasets=DATASETS_CONFIG_LIST,
    cudnn_benchmark=False,
    max_audio_len=22050 * 10,
    mixed_precision=False,
    test_sentences=[
        [
            "这是一个测试句子。",
            "SSB005",
            None,
            "zh",
        ],
        [
            "欢迎使用 YourTTS 模型。",
            "SSB009",
            None,
            "zh",
        ],
    ],
    use_weighted_sampler=True,
    weighted_sampler_attrs={"speaker_name": 1.0},
    weighted_sampler_multipliers={},
    speaker_encoder_loss_alpha=9.0,
)

# 加载数据样本
train_samples, eval_samples = load_tts_samples(
    config.datasets,
    eval_split=True,
    eval_split_max_size=config.eval_split_max_size,
    eval_split_size=config.eval_split_size,
    formatter=my_data_formatter,
)

print("train_samples",len(train_samples), "eval_samples",len(eval_samples))

print("Train Samples:", train_samples[:1])  # 打印前5个样本
print("Eval Samples:", eval_samples[:1])    # 打印前5个样本


# 初始化模型
model = Vits.init_from_config(config)

# 初始化训练器
trainer = Trainer(
    TrainerArgs(),
    config,
    output_path=OUT_PATH,
    model=model,
    train_samples=train_samples,
    eval_samples=eval_samples,
)

# 开始训练
trainer.fit()

def my_data_formatter(root_path, manifest_file, **kwargs):

    csv_file = os.path.join(root_path, manifest_file)
    items = []
    
    with open(csv_file, mode='r', encoding="utf-8") as file:
        reader = csv.DictReader(file)
        for row in reader:
            # 从 CSV 中读取每一列
            text = row["text"].strip()
            wav_path = row["audio_file"].strip()
            speaker_name = row["speaker_name"].strip()
            root_path = row["root_path"].strip()
            
            # 构造完整的音频文件路径
            wav_file = os.path.join(root_path, wav_path)
            
            # 不管文件是否存在，都添加到 items 中
            item = {
                "text": text,
                "audio_file": wav_file,
                "speaker_name": speaker_name,
                "root_path": root_path
            }
            items.append(item)
            
    return items