import numpy as np import pyworld from scipy.io import wavfile class FeatureInput(object): def __init__(self, samplerate=16000, hop_size=160): self.fs = samplerate self.hop = hop_size self.f0_bin = 256 self.f0_max = 1100.0 self.f0_min = 50.0 self.f0_mel_min = 1127 * np.log(1 + self.f0_min / 700) self.f0_mel_max = 1127 * np.log(1 + self.f0_max / 700) def compute_f0(self, audio, sr): x, sr = audio, self.fs assert sr == self.fs f0, t = pyworld.dio( x.astype(np.double), fs=sr, f0_ceil=800, frame_period=1000 * self.hop / sr, ) f0 = pyworld.stonemask(x.astype(np.double), f0, t, self.fs) for index, pitch in enumerate(f0): f0[index] = round(pitch, 1) return f0 # for numpy # code from diffsinger def coarse_f0(self, f0): f0_mel = 1127 * np.log(1 + f0 / 700) f0_mel[f0_mel > 0] = (f0_mel[f0_mel > 0] - self.f0_mel_min) * ( self.f0_bin - 2 ) / (self.f0_mel_max - self.f0_mel_min) + 1 # use 0 or 1 f0_mel[f0_mel <= 1] = 1 f0_mel[f0_mel > self.f0_bin - 1] = self.f0_bin - 1 f0_coarse = np.rint(f0_mel).astype(np.int) assert f0_coarse.max() <= 255 and f0_coarse.min() >= 1, ( f0_coarse.max(), f0_coarse.min(), ) return f0_coarse # for tensor # code from diffsinger def coarse_f0_ts(self, f0): f0_mel = 1127 * (1 + f0 / 700).log() f0_mel[f0_mel > 0] = (f0_mel[f0_mel > 0] - self.f0_mel_min) * ( self.f0_bin - 2 ) / (self.f0_mel_max - self.f0_mel_min) + 1 # use 0 or 1 f0_mel[f0_mel <= 1] = 1 f0_mel[f0_mel > self.f0_bin - 1] = self.f0_bin - 1 f0_coarse = (f0_mel + 0.5).long() assert f0_coarse.max() <= 255 and f0_coarse.min() >= 1, ( f0_coarse.max(), f0_coarse.min(), ) return f0_coarse def save_wav(self, wav, path): wav *= 32767 / max(0.01, np.max(np.abs(wav))) * 0.6 wavfile.write(path, self.fs, wav.astype(np.int16))