Method plot_wav

usr/diffspeech_task.py:112–121

(self, batch_idx, gt_wav, wav_out, is_mel=False, gt_f0=None, f0=None, name=None)

    # validation plots
    ############
    def plot_wav(self, batch_idx, gt_wav, wav_out, is_mel=False, gt_f0=None, f0=None, name=None):
        gt_wav = gt_wav[0].cpu().numpy()
        wav_out = wav_out[0].cpu().numpy()
        gt_f0 = gt_f0[0].cpu().numpy()
        f0 = f0[0].cpu().numpy()
        if is_mel:
            gt_wav = self.vocoder.spec2wav(gt_wav, f0=gt_f0)
            wav_out = self.vocoder.spec2wav(wav_out, f0=f0)
        self.logger.experiment.add_audio(f'gt_{batch_idx}', gt_wav, sample_rate=hparams['audio_sample_rate'], global_step=self.global_step)
        self.logger.experiment.add_audio(f'wav_{batch_idx}', wav_out, sample_rate=hparams['audio_sample_rate'], global_step=self.global_step)