synthesize_e2e.sh 中代码如下：
python3 ${BIN_DIR}/../synthesize.py \
    --am=tacotron2_aishell3 \
    --am_config=${config_path} \
    --am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
    --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 \
    --test_metadata=dump/test/norm/metadata.jsonl \
    --output_dir=${train_output_path}/test \
    --phones_dict=dump/phone_id_map.txt \
    --speaker_dict=dump/speaker_id_map.txt \
    --voice-cloning=True

if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
    # synthesize, vocoder is pwgan by default stage 0, stage 1 will use hifigan as vocoder
    CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh --stage 0 ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi

if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
    # synthesize_e2e, vocoder is pwgan by default stage 0, stage 1 will use hifigan as vocoder
    CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh --stage 0 ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi

`./local/synthesize.sh` calls `${BIN_DIR}/../synthesize.py`, which can synthesize waveform from `metadata.jsonl`. 

CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh --stage 0 ${conf_path} ${train_output_path} ${ckpt_name}

`--stage` controls the vocoder model during synthesis, which can be `0` or `1`, use `pwgan` or `hifigan` model as vocoder.

HiFiGAN checkpoint contains files listed below.
mb_melgan_csmsc_ckpt_0.1.1
├── default.yaml                    # default config used to train MultiBand MelGAN
├── feats_stats.npy                 # statistics used to normalize spectrogram when training MultiBand MelGAN
└── snapshot_iter_1000000.pdz       # generator parameters of MultiBand MelGAN

【报名】：1、3、2-3