使用預訓練模型執行語音客隆失敗

[arthurwu4work]

您好:

我根據aishell3/vc1 的說明下載預訓練模型並執行後，發生錯誤

AttributeError: n_mels

請問該如何解決

謝謝

=========================================================================

執行以下指令
CUDA_VISIBLE_DEVICES="0" ./local/voice_cloning.sh /media/rc3/MX500_1TB/TTS/wav_data/test_george2.wav

---

./local/voice_cloning.sh

#!/bin/bash

# config_path=$1
# train_output_path=$2
# ckpt_name=$3
# ge2e_params_path=$4
# ref_audio_dir=$5

ref_audio_dir=$1

python3 ${BIN_DIR}/../voice_cloning.py \
    --am=fastspeech2_aishell3 \
    --am_config=ge2e_ckpt_0.3/config.yaml \
    --am_ckpt=fastspeech2_nosil_aishell3_vc1_ckpt_0.5/snapshot_iter_96400.pdz \
    --am_stat=dump/train/speech_stats.npy \
    --voc=pwgan_aishell3 \
    --voc_config=pwg_aishell3_ckpt_0.5/default.yaml \
    --voc_ckpt=pwg_aishell3_ckpt_0.5/snapshot_iter_1000000.pdz \
    --voc_stat=pwg_aishell3_ckpt_0.5/feats_stats.npy \
    --ge2e_params_path=ge2e_ckpt_0.3/step-3000000.pdparams \
    --text="凯莫瑞安联合体的经济崩溃迫在眉睫。" \
    --input-dir=${ref_audio_dir} \
    --output-dir=/output \
    --phones-dict=fastspeech2_nosil_aishell3_vc1_ckpt_0.5/phone_id_map.txt

===============================================================================

log

/home/rc3/anaconda3/envs/py38_tts/lib/python3.8/site-packages/librosa/core/constantq.py:1059: DeprecationWarning: `np.complex` is a deprecated alias for the builtin `complex`. To silence this warning, use `complex` by itself. Doing this will not modify any behavior and is safe. If you specifically wanted the numpy scalar type, use `np.complex128` here.
Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations
  dtype=np.complex,
/home/rc3/.local/lib/python3.8/site-packages/_distutils_hack/__init__.py:35: UserWarning: Setuptools is replacing distutils.
  warnings.warn("Setuptools is replacing distutils.")
========Args========
am: fastspeech2_aishell3
am_ckpt: fastspeech2_nosil_aishell3_vc1_ckpt_0.5/snapshot_iter_96400.pdz
am_config: ge2e_ckpt_0.3/config.yaml
am_stat: dump/train/speech_stats.npy
ge2e_params_path: ge2e_ckpt_0.3/step-3000000.pdparams
input_dir: /media/rc3/MX500_1TB/TTS/wav_data/test_george2.wav
ngpu: 1
output_dir: ./output/
phones_dict: fastspeech2_nosil_aishell3_vc1_ckpt_0.5/phone_id_map.txt
text: "\u51EF\u83AB\u745E\u5B89\u8054\u5408\u4F53\u7684\u7ECF\u6D4E\u5D29\u6E83\u8FEB\
  \u5728\u7709\u776B\u3002"
use_ecapa: false
voc: pwgan_aishell3
voc_ckpt: pwg_aishell3_ckpt_0.5/snapshot_iter_1000000.pdz
voc_config: pwg_aishell3_ckpt_0.5/default.yaml
voc_stat: pwg_aishell3_ckpt_0.5/feats_stats.npy

========Config========
data:
  audio_norm_target_dBFS: -30
  mel_window_length: 25
  mel_window_step: 10
  min_pad_coverage: 0.75
  n_mels: 40
  partial_n_frames: 160
  partial_overlap_ratio: 0.5
  sampling_rate: 16000
  vad_max_silence_length: 6
  vad_moving_average_width: 8
  vad_window_length: 30
model:
  embedding_size: 256
  hidden_size: 256
  num_layers: 3
training:
  learning_rate_init: 0.0001
  max_iteration: 3000000
  save_interval: 10000
  speakers_per_batch: 64
  utterances_per_speaker: 10
  valid_interval: 10000
allow_cache: True
batch_max_steps: 24000
batch_size: 8
discriminator_grad_norm: 1
discriminator_optimizer_params:
  epsilon: 1e-06
  weight_decay: 0.0
discriminator_params:
  bias: True
  conv_channels: 64
  in_channels: 1
  kernel_size: 3
  layers: 10
  nonlinear_activation: LeakyReLU
  nonlinear_activation_params:
    negative_slope: 0.2
  out_channels: 1
  use_weight_norm: True
discriminator_scheduler_params:
  gamma: 0.5
  learning_rate: 5e-05
  step_size: 200000
discriminator_train_start_steps: 100000
eval_interval_steps: 1000
fmax: 7600
fmin: 80
fs: 24000
generator_grad_norm: 10
generator_optimizer_params:
  epsilon: 1e-06
  weight_decay: 0.0
generator_params:
  aux_channels: 80
  aux_context_window: 2
  dropout: 0.0
  gate_channels: 128
  in_channels: 1
  kernel_size: 3
  layers: 30
  out_channels: 1
  residual_channels: 64
  skip_channels: 64
  stacks: 3
  upsample_scales: [4, 5, 3, 5]
  use_weight_norm: True
generator_scheduler_params:
  gamma: 0.5
  learning_rate: 0.0001
  step_size: 200000
lambda_adv: 4.0
n_fft: 2048
n_mels: 80
n_shift: 300
num_save_intermediate_results: 4
num_snapshots: 10
num_workers: 4
pin_memory: True
remove_short_samples: True
save_interval_steps: 5000
seed: 42
stft_loss_params:
  fft_sizes: [1024, 2048, 512]
  hop_sizes: [120, 240, 50]
  win_lengths: [600, 1200, 240]
  window: hann
train_max_steps: 1000000
win_length: 1200
window: hann
Audio Processor Done!
W0505 14:03:14.293866 782957 gpu_resources.cc:61] Please NOTE: device: 0, GPU Compute Capability: 7.5, Driver API Version: 11.2, Runtime API Version: 10.2
W0505 14:03:14.296069 782957 gpu_resources.cc:91] device: 0, cuDNN Version: 8.1.
GE2E Done!
[2023-05-05 14:03:17,833] [    INFO] - Already cached /home/rc3/.paddlenlp/models/bert-base-chinese/bert-base-chinese-vocab.txt
[2023-05-05 14:03:17,840] [    INFO] - tokenizer config file saved in /home/rc3/.paddlenlp/models/bert-base-chinese/tokenizer_config.json
[2023-05-05 14:03:17,841] [    INFO] - Special tokens file saved in /home/rc3/.paddlenlp/models/bert-base-chinese/special_tokens_map.json
frontend done!
Building prefix dict from the default dictionary ...
[2023-05-05 14:03:17] [DEBUG] [__init__.py:113] Building prefix dict from the default dictionary ...
Loading model from cache /tmp/jieba.cache
[2023-05-05 14:03:17] [DEBUG] [__init__.py:132] Loading model from cache /tmp/jieba.cache
Loading model cost 0.478 seconds.
[2023-05-05 14:03:18] [DEBUG] [__init__.py:164] Loading model cost 0.478 seconds.
Prefix dict has been built successfully.
[2023-05-05 14:03:18] [DEBUG] [__init__.py:166] Prefix dict has been built successfully.
Traceback (most recent call last):
  File "/media/rc3/MX500_1TB/TTS/PaddleSpeech/paddlespeech/t2s/exps/fastspeech2/../voice_cloning.py", line 233, in <module>
    main()
  File "/media/rc3/MX500_1TB/TTS/PaddleSpeech/paddlespeech/t2s/exps/fastspeech2/../voice_cloning.py", line 229, in main
    voice_cloning(args)
  File "/media/rc3/MX500_1TB/TTS/PaddleSpeech/paddlespeech/t2s/exps/fastspeech2/../voice_cloning.py", line 101, in voice_cloning
    am_inference = get_am_inference(
  File "/media/rc3/MX500_1TB/TTS/PaddleSpeech/paddlespeech/t2s/exps/syn_utils.py", line 390, in get_am_inference
    odim = am_config.n_mels
  File "/home/rc3/anaconda3/envs/py38_tts/lib/python3.8/site-packages/yacs/config.py", line 141, in __getattr__
    raise AttributeError(name)
AttributeError: n_mels