diff options
Diffstat (limited to 'become_yukarin')
| -rw-r--r-- | become_yukarin/dataset/dataset.py | 6 | ||||
| -rw-r--r-- | become_yukarin/dataset/utility.py | 12 | ||||
| -rw-r--r-- | become_yukarin/param.py | 2 |
3 files changed, 9 insertions, 11 deletions
diff --git a/become_yukarin/dataset/dataset.py b/become_yukarin/dataset/dataset.py index b597bba..d259734 100644 --- a/become_yukarin/dataset/dataset.py +++ b/become_yukarin/dataset/dataset.py @@ -237,12 +237,12 @@ class DecodeFeatureProcess(BaseDataProcess): def __call__(self, data: numpy.ndarray, test): data = data.T - lens = [self._sizes[t] for t in self._targets] - assert data.shape[1] == sum(lens) + lasts = numpy.cumsum([self._sizes[t] for t in self._targets]).tolist() + assert data.shape[1] == lasts[-1] d = defaultdict(lambda: numpy.nan, **{ t: data[:, bef:aft] - for t, bef, aft in zip(self._targets, [0] + lens[:-1], lens) + for t, bef, aft in zip(self._targets, [0] + lasts[:-1], lasts) }) return AcousticFeature( f0=d['f0'], diff --git a/become_yukarin/dataset/utility.py b/become_yukarin/dataset/utility.py index c28b1df..9797818 100644 --- a/become_yukarin/dataset/utility.py +++ b/become_yukarin/dataset/utility.py @@ -13,8 +13,9 @@ class DTWAligner(object): assert x.ndim == 2 and y.ndim == 2 _, path = fastdtw.fastdtw(x, y, radius=radius, dist=dist) - self.normed_path_x = numpy.array(list(map(lambda l: l[0], path))) / len(x) - self.normed_path_y = numpy.array(list(map(lambda l: l[1], path))) / len(y) + path = numpy.array(path) + self.normed_path_x = path[:, 0] / len(x) + self.normed_path_y = path[:, 1] / len(y) def align_x(self, x): path = self._interp_path(self.normed_path_x, len(x)) @@ -34,10 +35,7 @@ class DTWAligner(object): @staticmethod def _interp_path(normed_path: numpy.ndarray, target_length: int): - base = numpy.linspace(0, 1, len(normed_path)) - target = numpy.linspace(0, 1, target_length) - path = scipy.interpolate.interp1d(base, normed_path)(target) - path = numpy.floor(path * target_length).astype(numpy.int) + path = numpy.floor(normed_path * target_length).astype(numpy.int) return path @@ -50,7 +48,7 @@ class MFCCAligner(DTWAligner): @classmethod def _calc_delta(cls, x): x = numpy.zeros_like(x, x.dtype) - x[:-1] = x[:-1] - x[1:] + x[:-1] = x[1:] - x[:-1] x[-1] = 0 return x diff --git a/become_yukarin/param.py b/become_yukarin/param.py index 0870bab..e6f46bc 100644 --- a/become_yukarin/param.py +++ b/become_yukarin/param.py @@ -9,7 +9,7 @@ class VoiceParam(NamedTuple): class AcousticFeatureParam(NamedTuple): frame_period: int = 5 - order: int = 59 + order: int = 25 alpha: float = 0.466 |