lt-tensor 0.0.1a34__py3-none-any.whl → 0.0.1a36__py3-none-any.whl

lt_tensor/processors/audio.py

@@ -73,7 +73,7 @@ class AudioProcessorConfig(ModelConfig):
     def post_process(self):
         self.n_stft = self.n_fft // 2 + 1
         # some functions needs this to be a non-zero or not None value.
-        self.f_min = max(self.f_min, (self.sample_rate / (self.n_fft - 1))  * 2)
+        self.f_min = max(self.f_min, (self.sample_rate / (self.n_fft - 1)) * 2)
         self.default_f_max = min(
             default(self.f_max, self.sample_rate // 2), self.sample_rate // 2
         )
@@ -105,7 +105,6 @@ class AudioProcessor(Model):
             f_min=self.cfg.f_min,
             f_max=self.cfg.f_max,
             mel_scale=self.cfg.mel_scale,
-            onesided=self.cfg.onesided,
             normalized=self.cfg.normalized,
         )
         self._mel_rscale = torchaudio.transforms.InverseMelScale(
@@ -122,17 +121,15 @@ class AudioProcessor(Model):
             (torch.hann_window(self.cfg.win_length) if window is None else window),
         )
 
-
-    
     def compute_mel(
         self,
         wave: Tensor,
-        raw_mel_only: bool = False,
         eps: float = 1e-5,
+        raw_mel_only: bool = False,
         *,
         _recall: bool = False,
     ) -> Tensor:
-        """Returns: [B, M, T]"""
+        """Returns: (M, T) or (B, M, T) if batched"""
         try:
             mel_tensor = self._mel_spec(wave.to(self.device))  # [M, T]
             if not raw_mel_only:
@@ -203,13 +200,16 @@ class AudioProcessor(Model):
             rms_ = []
             for i in range(B):
                 _t = _comp_rms_helper(i, audio, mel)
-                _r = librosa.feature.rms(**_t, **rms_kwargs)[
-                    0
-                ]
+                _r = librosa.feature.rms(**_t, **rms_kwargs)[0]
                 rms_.append(_r)
             return self.from_numpy_batch(rms_, default_device, default_dtype).squeeze()
 
-    def pitch_shift(self, audio: torch.Tensor, sample_rate: Optional[int] = None, n_steps: float = 2.0):
+    def pitch_shift(
+        self,
+        audio: torch.Tensor,
+        sample_rate: Optional[int] = None,
+        n_steps: float = 2.0,
+    ):
         """
         Shifts the pitch of an audio tensor by `n_steps` semitones.
 
@@ -225,21 +225,25 @@ class AudioProcessor(Model):
         src_dtype = audio.dtype
         audio = audio.squeeze()
         sample_rate = default(sample_rate, self.cfg.sample_rate)
+
         def _shift_one(wav):
             wav_np = self.to_numpy_safe(wav)
-            shifted_np = librosa.effects.pitch_shift(wav_np, sr=sample_rate, n_steps=n_steps)
+            shifted_np = librosa.effects.pitch_shift(
+                wav_np, sr=sample_rate, n_steps=n_steps
+            )
             return torch.from_numpy(shifted_np)
 
         if audio.ndim == 1:
             return _shift_one(audio).to(device=src_device, dtype=src_dtype)
-        return torch.stack([_shift_one(a) for a in audio]).to(device=src_device, dtype=src_dtype)
-
+        return torch.stack([_shift_one(a) for a in audio]).to(
+            device=src_device, dtype=src_dtype
+        )
 
     @staticmethod
-    def calc_pitch_fmin(sr:int, frame_length:float):
+    def calc_pitch_fmin(sr: int, frame_length: float):
         """For pitch f_min"""
         return (sr / (frame_length - 1)) * 2
-    
+
     def compute_pitch(
         self,
         audio: Tensor,
@@ -261,8 +265,10 @@ class AudioProcessor(Model):
             B = 1
         sr = default(sr, self.cfg.sample_rate)
         frame_length = default(frame_length, self.cfg.n_fft)
-        fmin = max(default(fmin, self.cfg.f_min), self.calc_pitch_fmin(sr, frame_length))
-        fmax = min(max(default(fmax, self.cfg.default_f_max), fmin+1), sr // 2)
+        fmin = max(
+            default(fmin, self.cfg.f_min), self.calc_pitch_fmin(sr, frame_length)
+        )
+        fmax = min(max(default(fmax, self.cfg.default_f_max), fmin + 1), sr // 2)
         hop_length = default(hop_length, self.cfg.hop_length)
         center = default(center, self.cfg.center)
         yn_kwargs = dict(
@@ -361,7 +367,7 @@ class AudioProcessor(Model):
         The modes available for upsampling are: `nearest`, `linear` (3D-only),
         `bilinear`, `bicubic` (4D-only), `trilinear` (5D-only)
         """
-
+        tensor = tensor.squeeze()
         if tensor.ndim == 2:  # [1, T]
             tensor = tensor.unsqueeze(1)  # [1, 1, T]
         elif tensor.ndim == 1:
@@ -384,7 +390,7 @@ class AudioProcessor(Model):
         hop_length: Optional[int] = None,
         win_length: Optional[int] = None,
         length: Optional[int] = None,
-        center: Optional[bool] = None,
+        center: bool = True,
         normalized: Optional[bool] = None,
         onesided: Optional[bool] = None,
         return_complex: bool = False,
@@ -403,7 +409,7 @@ class AudioProcessor(Model):
                 hop_length=default(hop_length, self.cfg.hop_length),
                 win_length=default(win_length, self.cfg.win_length),
                 window=window,
-                center=default(center, self.cfg.center),
+                center=center,
                 normalized=default(normalized, self.cfg.normalized),
                 onesided=default(onesided, self.cfg.onesided),
                 length=length,
@@ -421,44 +427,48 @@ class AudioProcessor(Model):
         self,
         wave: Tensor,
         length: Optional[int] = None,
-        *,
-        _recall: bool = False,
+        center: bool = True,
+        n_fft: Optional[int] = None,
+        hop_length: Optional[int] = None,
+        win_length: Optional[int] = None,
+        normalized: Optional[bool] = None,
+        onesided: Optional[bool] = None,
+        return_complex: bool = False,
     ):
-        try:
-            spectrogram = torch.stft(
-                input=wave,
-                n_fft=self.cfg.n_fft,
-                hop_length=self.cfg.hop_length,
-                win_length=self.cfg.win_length,
-                window=self.window,
-                center=self.cfg.center,
-                pad_mode="reflect",
-                normalized=self.cfg.normalized,
-                onesided=self.cfg.onesided,
-                return_complex=True,
-            )
-            return torch.istft(
-                spectrogram
-                * torch.full(
-                    spectrogram.size(),
-                    fill_value=1,
-                    device=spectrogram.device,
-                ),
-                n_fft=self.cfg.n_fft,
-                hop_length=self.cfg.hop_length,
-                win_length=self.cfg.win_length,
-                window=self.window,
-                length=length,
-                center=self.cfg.center,
-                normalized=self.cfg.normalized,
-                onesided=self.cfg.onesided,
-                return_complex=False,
-            )
-        except RuntimeError as e:
-            if not _recall and wave.device != self.window.device:
-                self.window = self.window.to(wave.device)
-                return self.istft_norm(wave, length, _recall=True)
-            raise e
+
+        if win_length is not None and win_length != self.cfg.win_length:
+            window = torch.hann_window(win_length, device=wave.device)
+        else:
+            window = self.window
+        spectrogram = torch.stft(
+            input=wave,
+            n_fft=default(n_fft, self.cfg.n_fft),
+            hop_length=default(hop_length, self.cfg.hop_length),
+            win_length=default(win_length, self.cfg.win_length),
+            window=window,
+            center=center,
+            pad_mode="reflect",
+            normalized=default(normalized, self.cfg.normalized),
+            onesided=default(onesided, self.cfg.onesided),
+            return_complex=True,
+        )
+        return torch.istft(
+            spectrogram
+            * torch.full(
+                spectrogram.size(),
+                fill_value=1,
+                device=spectrogram.device,
+            ),
+            n_fft=default(n_fft, self.cfg.n_fft),
+            hop_length=default(hop_length, self.cfg.hop_length),
+            win_length=default(win_length, self.cfg.win_length),
+            window=self.window,
+            length=length,
+            center=center,
+            normalized=default(normalized, self.cfg.normalized),
+            onesided=default(onesided, self.cfg.onesided),
+            return_complex=return_complex,
+        )
 
     def load_audio(
         self,

{lt_tensor-0.0.1a34.dist-info → lt_tensor-0.0.1a36.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lt-tensor
-Version: 0.0.1a34
+Version: 0.0.1a36
 Summary: General utilities for PyTorch and others. Built for general use.
 Home-page: https://github.com/gr1336/lt-tensor/
 Author: gr1336

lt_tensor-0.0.1a36.dist-info/RECORD

@@ -0,0 +1,43 @@
+lt_tensor/__init__.py,sha256=nBbiGH1byHU0aTTKKorRj8MIEO2oEMBXl7kt5DOCatU,441
+lt_tensor/config_templates.py,sha256=F9UvL8paAjkSvio890kp8WznpYeI50pYnm9iqQroBxk,2797
+lt_tensor/losses.py,sha256=Heco_WyoC1HkNkcJEircOAzS9umusATHiNAG-FKGyzc,8918
+lt_tensor/lr_schedulers.py,sha256=6_vcfaPHrozfH3wvmNEdKSFYl6iTIijYoHL8vuG-45U,7651
+lt_tensor/math_ops.py,sha256=ahX6Z1Mt3X-FhmwSZYZea5mB1B0S8GDuvKPfAm5e_FQ,2646
+lt_tensor/misc_utils.py,sha256=stL6q3M7S2N4FBICFYbgYpdPDrJRlwmr24-iCXMRifM,28933
+lt_tensor/model_base.py,sha256=5T4dbAh4MXbQmPRpihGtMYwTY8sJTQOhY6An3VboM58,18086
+lt_tensor/monotonic_align.py,sha256=LhBd8p1xdBzg6jQrQX1j7b4PNeYGwIqM24zcU-pHOLE,2239
+lt_tensor/noise_tools.py,sha256=wFeAsHhLhSlEc5XU5LbFKaXoHeVxrWjiMeljjGdIKyM,11363
+lt_tensor/torch_commons.py,sha256=8l0bxmrAzwvyqjivCIVISXlbvKarlg4DdE0BOGSnMuQ,812
+lt_tensor/transform.py,sha256=dZm8T_ov0blHMQu6nGiehsdG1VSB7bZBUVmTkT-PBdc,13257
+lt_tensor/model_zoo/__init__.py,sha256=yPUVchgVhU2nAJ2ocA4HFfG7IMEiBu8qOi8I1KWTTkU,404
+lt_tensor/model_zoo/basic.py,sha256=pI8HyiHK-cmWcEEaVY_EduUJOjZW6HOtXvJd8Rbhq30,15452
+lt_tensor/model_zoo/convs.py,sha256=Tws0jrPfs9m7OLmJ30W0AfkAvZgppW7lNi4xt0e-qRU,3518
+lt_tensor/model_zoo/features.py,sha256=DO8dlE0kmPKTNC1Xkv9wKegOOYkQa_rkxM4hhcNwJWA,15655
+lt_tensor/model_zoo/fusion.py,sha256=usC1bcjQRNivDc8xzkIS5T1glm78OLcs2V_tPqfp-eI,5422
+lt_tensor/model_zoo/pos_encoder.py,sha256=3d1EYLinCU9UAy-WuEWeYMGhMqaGknCiQ5qEmhw_UYM,4487
+lt_tensor/model_zoo/residual.py,sha256=tMXgif9Ggep9bk75K93yueeU5vk5S25AGCRFwOQOyB8,6452
+lt_tensor/model_zoo/transformer.py,sha256=HUFoFFh7EQJErxdd9XIxhssdjvNVx2tNGDJOTUfwG2A,4301
+lt_tensor/model_zoo/activations/__init__.py,sha256=f_IsuC-SaFsX6w4OtBWa5bbS4TqR90X-cvLxGUgYfjk,67
+lt_tensor/model_zoo/activations/alias_free/__init__.py,sha256=dgLjatRm9nusoPVOl1pvCef5rZsaRfS3BJUs05SPYzw,64
+lt_tensor/model_zoo/activations/alias_free/act.py,sha256=1wxmab2kMD88L6wsQgf3t25dBwR7_he2eM1DlV0FQak,1424
+lt_tensor/model_zoo/activations/alias_free/filter.py,sha256=5TvXESv31toD5sePBe_OUJJfMXv6Ohwmx2YawjQL-pk,6004
+lt_tensor/model_zoo/activations/alias_free/resample.py,sha256=3iM4fNr9fLNXXMyXvzW-MwkSjOZOrMZLfS80UHs6zk0,3386
+lt_tensor/model_zoo/activations/snake/__init__.py,sha256=AtOAbJuMinxmKkppITGMzRbcbPQaALnl9mCtl1c3x0Q,4356
+lt_tensor/model_zoo/audio_models/__init__.py,sha256=WwiP9MekJreMOfKPWLl24VkRJIpLk6hhL8ch0aKgOss,103
+lt_tensor/model_zoo/audio_models/resblocks.py,sha256=u-foHxaFDUICjxSkpyHXljQYQG9zMxVYaOGqLR_nJ-k,7978
+lt_tensor/model_zoo/audio_models/bigvgan/__init__.py,sha256=4EZG8Non75dHoDCizMHbMTvPrKwdUlPYGHc7hkfT_nw,8526
+lt_tensor/model_zoo/audio_models/diffwave/__init__.py,sha256=PDuDYN1omD1RoAXcmxH3tEgfAuM3ZHAWzimD6ElMqEQ,9073
+lt_tensor/model_zoo/audio_models/hifigan/__init__.py,sha256=ITSXHg3c0Um1P2HaPaXkQKI7meG5Ne60wTbyyYju3hY,6360
+lt_tensor/model_zoo/audio_models/istft/__init__.py,sha256=blICjLX_z_IFmR3_TCz_dJiSayLYGza9eG6fd9aKyvE,7448
+lt_tensor/model_zoo/losses/__init__.py,sha256=B9RAUxBiOZwooztnij1oLeRwZ7_MjnN3mPoum7saD6s,59
+lt_tensor/model_zoo/losses/discriminators.py,sha256=o4cicNdOv0jH3ink7jTNeDqOnwmkmRtEj9E7IUIGnEI,31866
+lt_tensor/model_zoo/losses/CQT/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+lt_tensor/model_zoo/losses/CQT/transforms.py,sha256=Vkid0J9dqLnlINfyyUlQf-qB3gOQAgU7W9j7xLOjDFw,13218
+lt_tensor/model_zoo/losses/CQT/utils.py,sha256=twGw6FVD7V5Ksfx_1BUEN3EP1tAS6wo-9LL3VnuHB8c,16751
+lt_tensor/processors/__init__.py,sha256=Pvxhh0KR65zLCgUd53_k5Z0y5JWWcO0ZBXFK9rv0o5w,109
+lt_tensor/processors/audio.py,sha256=3YzyEpMwh124rb1KMAly62qweeruF200BnM-vQIbzy0,18645
+lt_tensor-0.0.1a36.dist-info/licenses/LICENSE,sha256=TbiyJWLgNqqgqhfCnrGwFIxy7EqGNrIZZcKhHrefcuU,11354
+lt_tensor-0.0.1a36.dist-info/METADATA,sha256=mTmnoWn8EG48j_VOM3rr_8RLLgaxB5pWZE1tkPdFrac,1062
+lt_tensor-0.0.1a36.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+lt_tensor-0.0.1a36.dist-info/top_level.txt,sha256=35FuhFeXnUyvHWdbVHGPh0hS8euofafnJ_GJAVSF4Kk,10
+lt_tensor-0.0.1a36.dist-info/RECORD,,

lt_tensor/model_zoo/activations/alias_free_torch/__init__.py

@@ -1 +0,0 @@
-from . import *

lt_tensor-0.0.1a34.dist-info/RECORD

@@ -1,37 +0,0 @@
-lt_tensor/__init__.py,sha256=WAGPuMPq5c4DGAJ57x1Ykgzg3vMlLq9BiWk5EdJcUsU,441
-lt_tensor/config_templates.py,sha256=F9UvL8paAjkSvio890kp8WznpYeI50pYnm9iqQroBxk,2797
-lt_tensor/losses.py,sha256=fHVMqOFo3ekjORYy89R_aRjmtT6lo27Z1egzOYjQ1W8,8646
-lt_tensor/lr_schedulers.py,sha256=LSZzqrOOLzSthD8k-W4cYPJt0vCjmHkiJkLr5e3yRTE,3659
-lt_tensor/math_ops.py,sha256=ahX6Z1Mt3X-FhmwSZYZea5mB1B0S8GDuvKPfAm5e_FQ,2646
-lt_tensor/misc_utils.py,sha256=N2r3UmxC4RM2BZBQhpjDZ_BKLrzsyIlKzopTzJbnjFU,28962
-lt_tensor/model_base.py,sha256=5T4dbAh4MXbQmPRpihGtMYwTY8sJTQOhY6An3VboM58,18086
-lt_tensor/monotonic_align.py,sha256=LhBd8p1xdBzg6jQrQX1j7b4PNeYGwIqM24zcU-pHOLE,2239
-lt_tensor/noise_tools.py,sha256=wFeAsHhLhSlEc5XU5LbFKaXoHeVxrWjiMeljjGdIKyM,11363
-lt_tensor/torch_commons.py,sha256=8l0bxmrAzwvyqjivCIVISXlbvKarlg4DdE0BOGSnMuQ,812
-lt_tensor/transform.py,sha256=dZm8T_ov0blHMQu6nGiehsdG1VSB7bZBUVmTkT-PBdc,13257
-lt_tensor/model_zoo/__init__.py,sha256=yPUVchgVhU2nAJ2ocA4HFfG7IMEiBu8qOi8I1KWTTkU,404
-lt_tensor/model_zoo/basic.py,sha256=pI8HyiHK-cmWcEEaVY_EduUJOjZW6HOtXvJd8Rbhq30,15452
-lt_tensor/model_zoo/convs.py,sha256=YQRxek75Qpsha8nfc7wLhmJS9XxPeCa4WxuftLg6IcE,3927
-lt_tensor/model_zoo/features.py,sha256=DO8dlE0kmPKTNC1Xkv9wKegOOYkQa_rkxM4hhcNwJWA,15655
-lt_tensor/model_zoo/fusion.py,sha256=usC1bcjQRNivDc8xzkIS5T1glm78OLcs2V_tPqfp-eI,5422
-lt_tensor/model_zoo/pos_encoder.py,sha256=3d1EYLinCU9UAy-WuEWeYMGhMqaGknCiQ5qEmhw_UYM,4487
-lt_tensor/model_zoo/residual.py,sha256=tMXgif9Ggep9bk75K93yueeU5vk5S25AGCRFwOQOyB8,6452
-lt_tensor/model_zoo/transformer.py,sha256=HUFoFFh7EQJErxdd9XIxhssdjvNVx2tNGDJOTUfwG2A,4301
-lt_tensor/model_zoo/activations/alias_free_torch/__init__.py,sha256=ovguP4wzQEDNguczwiZnhMm4dRRVcvnzmHrfQtlRCNQ,15
-lt_tensor/model_zoo/activations/alias_free_torch/act.py,sha256=h79C93GzbSrCq4ui6iO7DjJLuJ7QK_ag_TU-WAcj0NI,1405
-lt_tensor/model_zoo/activations/alias_free_torch/filter.py,sha256=5TvXESv31toD5sePBe_OUJJfMXv6Ohwmx2YawjQL-pk,6004
-lt_tensor/model_zoo/activations/alias_free_torch/resample.py,sha256=3iM4fNr9fLNXXMyXvzW-MwkSjOZOrMZLfS80UHs6zk0,3386
-lt_tensor/model_zoo/activations/snake/__init__.py,sha256=Adb_xe-7YdYsNxvlSSO9zkae-cu7ElxkBKE3trDtOus,4517
-lt_tensor/model_zoo/audio_models/__init__.py,sha256=MoG9YjxLyvscq_6njK1ljGBletK9iedBXt66bplzW-s,83
-lt_tensor/model_zoo/audio_models/diffwave/__init__.py,sha256=PDuDYN1omD1RoAXcmxH3tEgfAuM3ZHAWzimD6ElMqEQ,9073
-lt_tensor/model_zoo/audio_models/hifigan/__init__.py,sha256=7GJqKLw7-juXpfp5IFzjASLut0uouDhjZ1CQknf3H68,16533
-lt_tensor/model_zoo/audio_models/istft/__init__.py,sha256=ltIuD9t1gmS3bTmCqZIwJHKrhC6DYya3OaXlskWX9kw,17606
-lt_tensor/model_zoo/losses/__init__.py,sha256=B9RAUxBiOZwooztnij1oLeRwZ7_MjnN3mPoum7saD6s,59
-lt_tensor/model_zoo/losses/discriminators.py,sha256=ZpyByFgc7L7uV_XRBsV9vkdVItbJO3z--Y6LlvTvtwY,20765
-lt_tensor/processors/__init__.py,sha256=Pvxhh0KR65zLCgUd53_k5Z0y5JWWcO0ZBXFK9rv0o5w,109
-lt_tensor/processors/audio.py,sha256=HNr1GS-6M2q0Rda4cErf5y2Jlc9f4jD58FvpX2ua9d4,18369
-lt_tensor-0.0.1a34.dist-info/licenses/LICENSE,sha256=TbiyJWLgNqqgqhfCnrGwFIxy7EqGNrIZZcKhHrefcuU,11354
-lt_tensor-0.0.1a34.dist-info/METADATA,sha256=WkTafcY5nYZbrZ7WzUc3JXnmg9NtUAXrchx42dCok9I,1062
-lt_tensor-0.0.1a34.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-lt_tensor-0.0.1a34.dist-info/top_level.txt,sha256=35FuhFeXnUyvHWdbVHGPh0hS8euofafnJ_GJAVSF4Kk,10
-lt_tensor-0.0.1a34.dist-info/RECORD,,