storytold/storyteller-ml
2
0
Fork 0
mirror of https://github.com/storytold/storyteller-ml.git synced 2026-10-09 00:09:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

lots more logging

Browse Source
This commit is contained in:
Brandon Thomas
2023-05-02 02:24:05 -07:00
parent 5a42c33015
commit 7d8b2c8308
1 changed files with 39 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
voice_conversion/so-vits-svc-paul/src/so_vits_svc_fork/inference/infer_tool.py
+39
View File
@@ -165,33 +165,72 @@ class Svc:
"crepe", "crepe-tiny", "parselmouth", "dio", "harvest"
] = "dio",
):
LOG.warn("get_unit_f0(): f0_method = {}".format(f0_method))
LOG.warn("get_unit_f0(): compute_f0() ...")
f0 = utils.compute_f0(
audio,
sampling_rate=self.target_sample,
hop_length=self.hop_size,
method=f0_method,
)
LOG.warn("get_unit_f0(): compute_f0() - END")
LOG.warn("get_unit_f0(): interpolate_f0()... ")
f0, uv = utils.interpolate_f0(f0)
LOG.warn("get_unit_f0(): interpolate_f0() - END")
f0 = torch.FloatTensor(f0)
uv = torch.FloatTensor(uv)
f0 = f0 * 2 ** (tran / 12)
f0 = f0.unsqueeze(0).to(self.dev)
uv = uv.unsqueeze(0).to(self.dev)
LOG.warn("get_unit_f0(): librosa.resample() ...")
wav16k = librosa.resample(
audio, orig_sr=self.target_sample, target_sr=HUBERT_SAMPLING_RATE
)
LOG.warn("get_unit_f0(): librosa.resample() - END")
LOG.warn("get_unit_f0(): torch.from_numpy()... ")
wav16k = torch.from_numpy(wav16k).to(self.dev)
LOG.warn("get_unit_f0(): torch.from_numpy() - END ")
LOG.warn("get_unit_f0(): utils.get_hubert_contet() ...")
c = utils.get_hubert_content(self.hubert_model, wav_16k_tensor=wav16k)
LOG.warn("get_unit_f0(): utils.get_hubert_contet() - END")
LOG.warn("get_unit_f0(): utils.repeat_expand_2d() ... ")
c = utils.repeat_expand_2d(c.squeeze(0), f0.shape[1])
LOG.warn("get_unit_f0(): utils.repeat_expand_2d() - DONE")
if cluster_infer_ratio != 0:
LOG.warn("get_unit_f0(): cluster.get_cluster_center_result() ...")
cluster_c = cluster.get_cluster_center_result(
self.cluster_model, c.cpu().numpy().T, speaker
).T
LOG.warn("get_unit_f0(): cluster.get_cluster_center_result() - END")
cluster_c = torch.FloatTensor(cluster_c).to(self.dev)
c = cluster_infer_ratio * cluster_c + (1 - cluster_infer_ratio) * c
LOG.warn("get_unit_f0(): cluster.get_cluster_center_result() - MATH DONE")
LOG.warn("get_unit_f0(): c.unsqueze()")
c = c.unsqueeze(0)
return c, f0, uv