Fix lint: ruff E711

Nugine · Nugine · commit 0f37f0cb80ac · 2024-05-04T15:54:13.000+08:00
diff --git a/evaluation/features/long_term_average_spectrum.py b/evaluation/features/long_term_average_spectrum.py
@@ -9,7 +9,7 @@
 
 def extract_ltas(audio, fs=None, n_fft=1024, hop_length=256):
     """Extract Long-Term Average Spectrum for a given audio."""
-    if fs != None:
+    if fs is not None:
         y, _ = librosa.load(audio, sr=fs)
     else:
         y, fs = librosa.load(audio)
diff --git a/evaluation/features/signal_to_noise_ratio.py b/evaluation/features/signal_to_noise_ratio.py
@@ -79,7 +79,7 @@ def getHarmonics(fund, sr, nHarmonics=6, aliased=False):
 
 def extract_snr(audio, sr=None):
     """Extract Signal-to-Noise Ratio for a given audio."""
-    if sr != None:
+    if sr is not None:
         audio, _ = librosa.load(audio, sr=sr)
     else:
         audio, sr = librosa.load(audio, sr=sr)
diff --git a/evaluation/features/singing_power_ratio.py b/evaluation/features/singing_power_ratio.py
@@ -37,7 +37,7 @@ def extract_spr(
     pitch_min: lower limit for f0 quantization.
     """
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio, _ = librosa.load(audio, sr=fs)
     else:
         audio, fs = librosa.load(audio)
diff --git a/evaluation/metrics/energy/energy_pearson_coefficients.py b/evaluation/metrics/energy/energy_pearson_coefficients.py
@@ -42,7 +42,7 @@ def extract_energy_pearson_coeffcients(
     pearson = PearsonCorrCoef()
 
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/energy/energy_rmse.py b/evaluation/metrics/energy/energy_rmse.py
@@ -37,7 +37,7 @@ def extract_energy_rmse(
     db_scale = kwargs["db_scale"]
 
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/f0/f0_pearson_coefficients.py b/evaluation/metrics/f0/f0_pearson_coefficients.py
@@ -46,7 +46,7 @@ def extract_fpc(
     pearson = PearsonCorrCoef()
 
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/f0/f0_periodicity_rmse.py b/evaluation/metrics/f0/f0_periodicity_rmse.py
@@ -31,7 +31,7 @@ def extract_f0_periodicity_rmse(
     method = kwargs["method"]
 
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/f0/f0_rmse.py b/evaluation/metrics/f0/f0_rmse.py
@@ -45,7 +45,7 @@ def extract_f0rmse(
     need_mean = kwargs["need_mean"]
 
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/f0/v_uv_f1.py b/evaluation/metrics/f0/v_uv_f1.py
@@ -44,7 +44,7 @@ def extract_f1_v_uv(
     method = kwargs["method"]
 
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/spectrogram/mel_cepstral_distortion.py b/evaluation/metrics/spectrogram/mel_cepstral_distortion.py
@@ -17,7 +17,7 @@ def extract_mcd(audio_ref, audio_deg, **kwargs):
     fs = kwargs["fs"]
 
     mcd_toolbox = Calculate_MCD(MCD_mode="dtw_sl")
-    if fs != None:
+    if fs is not None:
         mcd_toolbox.SAMPLING_RATE = fs
     mcd_value = mcd_toolbox.calculate_mcd(audio_ref, audio_deg)
 
diff --git a/evaluation/metrics/spectrogram/multi_resolution_stft_distance.py b/evaluation/metrics/spectrogram/multi_resolution_stft_distance.py
@@ -29,7 +29,7 @@ def extract_mstft(
     method = kwargs["method"]
 
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/spectrogram/pesq.py b/evaluation/metrics/spectrogram/pesq.py
@@ -24,7 +24,7 @@ def extract_pesq(audio_ref, audio_deg, **kwargs):
     method = kwargs["method"]
 
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/spectrogram/scale_invariant_signal_to_distortion_ratio.py b/evaluation/metrics/spectrogram/scale_invariant_signal_to_distortion_ratio.py
@@ -19,7 +19,7 @@ def extract_si_sdr(audio_ref, audio_deg, **kwargs):
 
     si_sdr = ScaleInvariantSignalDistortionRatio()
 
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/spectrogram/scale_invariant_signal_to_noise_ratio.py b/evaluation/metrics/spectrogram/scale_invariant_signal_to_noise_ratio.py
@@ -19,7 +19,7 @@ def extract_si_snr(audio_ref, audio_deg, **kwargs):
 
     si_snr = ScaleInvariantSignalNoiseRatio()
 
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/evaluation/metrics/spectrogram/short_time_objective_intelligibility.py b/evaluation/metrics/spectrogram/short_time_objective_intelligibility.py
@@ -25,7 +25,7 @@ def extract_stoi(audio_ref, audio_deg, **kwargs):
     method = kwargs["method"]
 
     # Load audio
-    if fs != None:
+    if fs is not None:
         audio_ref, _ = librosa.load(audio_ref, sr=fs)
         audio_deg, _ = librosa.load(audio_deg, sr=fs)
     else:
diff --git a/models/codec/ns3_codec/transformer.py b/models/codec/ns3_codec/transformer.py
@@ -129,7 +129,7 @@ def forward(self, x, key_padding_mask, conditon=None):
         else:
             x = self.ln_1(x)
 
-        if key_padding_mask != None:
+        if key_padding_mask is not None:
             key_padding_mask_input = ~(key_padding_mask.bool())
         else:
             key_padding_mask_input = None
@@ -186,7 +186,7 @@ def __init__(
         )
         self.use_cln = use_cln if use_cln is not None else cfg.use_cln
 
-        if enc_emb_tokens != None:
+        if enc_emb_tokens is not None:
             self.use_enc_emb = True
             self.enc_emb_tokens = enc_emb_tokens
         else:
diff --git a/models/tta/autoencoder/autoencoder_trainer.py b/models/tta/autoencoder/autoencoder_trainer.py
@@ -91,7 +91,7 @@ def build_criterion(self):
         return AutoencoderLossWithDiscriminator(self.cfg.model.loss)
 
     def get_state_dict(self):
-        if self.scheduler != None:
+        if self.scheduler is not None:
             state_dict = {
                 "model": self.model.state_dict(),
                 "optimizer_ae": self.optimizer["opt_ae"].state_dict(),
@@ -119,7 +119,7 @@ def load_model(self, checkpoint):
         self.model.load_state_dict(checkpoint["model"])
         self.optimizer["opt_ae"].load_state_dict(checkpoint["optimizer_ae"])
         self.optimizer["opt_disc"].load_state_dict(checkpoint["optimizer_disc"])
-        if self.scheduler != None:
+        if self.scheduler is not None:
             self.scheduler.load_state_dict(checkpoint["scheduler"])
 
     def build_model(self):
diff --git a/models/tta/ldm/audioldm_trainer.py b/models/tta/ldm/audioldm_trainer.py
@@ -122,7 +122,7 @@ def build_criterion(self):
         return criterion
 
     def get_state_dict(self):
-        if self.scheduler != None:
+        if self.scheduler is not None:
             state_dict = {
                 "model": self.model.state_dict(),
                 "optimizer": self.optimizer.state_dict(),
@@ -147,7 +147,7 @@ def load_model(self, checkpoint):
 
         self.model.load_state_dict(checkpoint["model"])
         self.optimizer.load_state_dict(checkpoint["optimizer"])
-        if self.scheduler != None:
+        if self.scheduler is not None:
             self.scheduler.load_state_dict(checkpoint["scheduler"])
 
     def build_model(self):
diff --git a/models/tts/naturalspeech2/wavenet.py b/models/tts/naturalspeech2/wavenet.py
@@ -98,7 +98,7 @@ def forward(self, x, x_mask, cond, diffusion_step, spk_query_emb):
         cond = self.cond_proj(cond)  # (B, 2*d, T)
 
         y = x + diffusion_step
-        if x_mask != None:
+        if x_mask is not None:
             y = y * x_mask.to(y.dtype)[:, None, :]  # (B, 2*d, T)
 
         if self.has_cattn:
@@ -120,7 +120,7 @@ def forward(self, x, x_mask, cond, diffusion_step, spk_query_emb):
 
         residual, skip = torch.chunk(y, 2, dim=1)
 
-        if x_mask != None:
+        if x_mask is not None:
             residual = residual * x_mask.to(y.dtype)[:, None, :]
             skip = skip * x_mask.to(y.dtype)[:, None, :]
 
diff --git a/models/vocoders/diffusion/diffusion_vocoder_inference.py b/models/vocoders/diffusion/diffusion_vocoder_inference.py
@@ -87,10 +87,10 @@ def synthesis_audios(cfg, model, mels, f0s=None, batch_size=None, fast_inference
 
     # Pad the given list into tensors
     mel_batches, mel_frames = pad_mels_to_tensors(mels, batch_size)
-    if f0s != None:
+    if f0s is not None:
         f0_batches = pad_f0_to_tensors(f0s, batch_size)
 
-    if f0s == None:
+    if f0s is None:
         for mel_batch, mel_frame in zip(mel_batches, mel_frames):
             for i in range(mel_batch.shape[0]):
                 mel = mel_batch[i]
diff --git a/models/vocoders/gan/gan_vocoder_inference.py b/models/vocoders/gan/gan_vocoder_inference.py
@@ -19,10 +19,10 @@ def vocoder_inference(cfg, model, mels, f0s=None, device=None, fast_inference=Fa
 
     with torch.no_grad():
         mels = mels.to(device)
-        if f0s != None:
+        if f0s is not None:
             f0s = f0s.to(device)
 
-        if f0s == None and not cfg.preprocess.extract_amplitude_phase:
+        if f0s is None and not cfg.preprocess.extract_amplitude_phase:
             output = model.forward(mels)
         elif cfg.preprocess.extract_amplitude_phase:
             (
@@ -52,10 +52,10 @@ def synthesis_audios(cfg, model, mels, f0s=None, batch_size=None, fast_inference
 
     # Pad the given list into tensors
     mel_batches, mel_frames = pad_mels_to_tensors(mels, batch_size)
-    if f0s != None:
+    if f0s is not None:
         f0_batches = pad_f0_to_tensors(f0s, batch_size)
 
-    if f0s == None:
+    if f0s is None:
         for mel_batch, mel_frame in zip(mel_batches, mel_frames):
             for i in range(mel_batch.shape[0]):
                 mel = mel_batch[i]
diff --git a/modules/naturalpseech2/transformers.py b/modules/naturalpseech2/transformers.py
@@ -130,7 +130,7 @@ def forward(self, x, key_padding_mask, conditon=None):
         else:
             x = self.ln_1(x)
 
-        if key_padding_mask != None:
+        if key_padding_mask is not None:
             key_padding_mask_input = ~(key_padding_mask.bool())
         else:
             key_padding_mask_input = None
@@ -187,7 +187,7 @@ def __init__(
         )
         self.use_cln = use_cln if use_cln is not None else cfg.use_cln
 
-        if enc_emb_tokens != None:
+        if enc_emb_tokens is not None:
             self.use_enc_emb = True
             self.enc_emb_tokens = enc_emb_tokens
         else:
diff --git a/preprocessors/ljspeech.py b/preprocessors/ljspeech.py
@@ -98,9 +98,9 @@ def get_uid2utt(ljspeech_path, dataset, cfg):
 def split_dataset(
     lines, test_rate=0.05, valid_rate=0.05, test_size=None, valid_size=None
 ):
-    if test_size == None:
+    if test_size is None:
         test_size = int(len(lines) * test_rate)
-    if valid_size == None:
+    if valid_size is None:
         valid_size = int(len(lines) * valid_rate)
     random.shuffle(lines)
 
diff --git a/processors/content_extractor.py b/processors/content_extractor.py
@@ -249,7 +249,7 @@ def save_feature(self, utt, content_feature):
             content_feature (tensor): content feature of one utterance
         """
         uid = utt["Uid"]
-        assert self.extractor_type != None
+        assert self.extractor_type is not None
         out_dir = os.path.join(
             self.cfg.preprocess.processed_dir, utt["Dataset"], self.extractor_type
         )
@@ -315,7 +315,7 @@ def __init__(self, cfg):
         super(ContentvecExtractor, self).__init__(cfg, extractor_type="contentvec")
 
     def load_model(self):
-        assert self.model == None
+        assert self.model is None
         # Load model
         ckpt_path = self.cfg.preprocess.contentvec_file
         print("Load Contentvec Model...")
@@ -445,8 +445,8 @@ def __init__(self, cfg):
         self.preprocessor = None
 
     def load_model(self):
-        assert self.model == None
-        assert self.preprocessor == None
+        assert self.model is None
+        assert self.preprocessor is None
 
         print("Loading MERT Model: ...", self.cfg.preprocess.mert_model)
 
diff --git a/utils/util.py b/utils/util.py
@@ -62,7 +62,7 @@ def pad_f0_to_tensors(f0s, batched=None):
     # Initialize
     tensors = []
 
-    if batched == None:
+    if batched is None:
         # Get the max frame for padding
         size = -1
         for f0 in f0s:
@@ -124,7 +124,7 @@ def pad_mels_to_tensors(mels, batched=None):
     mel_frames = []
 
     # Split mel-specs into batches to avoid cuda memory exceed
-    if batched == None:
+    if batched is None:
         # Get the max frame for padding
         size = -1
         for mel in mels:
