some change precision audio processing (#94)

* some change precision audio processing

* fix clipping problem in resample

resample sometimes causes signal clipping, not just librosa.resample

* fix error

extract_f0_print.py

@@ -33,7 +33,9 @@ class FeatureInput(object):
         self.f0_mel_max = 1127 * np.log(1 + self.f0_max / 700)
 
     def compute_f0(self, path, f0_method):
-        x, sr = librosa.load(path, self.fs)
+        # default resample type of librosa.resample is "soxr_hq".
+        # Quality: soxr_vhq > soxr_hq
+        x, sr = librosa.load(path, self.fs, res_type='soxr_vhq')
         p_len = x.shape[0] // self.hop
         assert sr == self.fs
         if f0_method == "pm":

my_utils.py

@@ -12,10 +12,10 @@ def load_audio(file, sr):
         )  # 防止小白拷路径头尾带了空格和"和回车
         out, _ = (
             ffmpeg.input(file, threads=0)
-            .output("-", format="s16le", acodec="pcm_s16le", ac=1, ar=sr)
+            .output("-", format="f32le", acodec="pcm_f32le", ac=1, ar=sr)
             .run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True)
         )
     except Exception as e:
         raise RuntimeError(f"Failed to load audio: {e}")
 
-    return np.frombuffer(out, np.int16).flatten().astype(np.float32) / 32768.0
+    return np.frombuffer(out, np.float32).flatten()

train/data_utils.py

@@ -98,7 +98,10 @@ class TextAudioLoaderMultiNSFsid(torch.utils.data.Dataset):
                     sampling_rate, self.sampling_rate
                 )
             )
-        audio_norm = audio / self.max_wav_value
+        audio_norm = audio
+#        audio_norm = audio / self.max_wav_value
+#        audio_norm = audio / np.abs(audio).max()
+
         audio_norm = audio_norm.unsqueeze(0)
         spec_filename = filename.replace(".wav", ".spec.pt")
         if os.path.exists(spec_filename):
@@ -287,7 +290,10 @@ class TextAudioLoader(torch.utils.data.Dataset):
                     sampling_rate, self.sampling_rate
                 )
             )
-        audio_norm = audio / self.max_wav_value
+        audio_norm = audio
+#        audio_norm = audio / self.max_wav_value
+#        audio_norm = audio / np.abs(audio).max()
+
         audio_norm = audio_norm.unsqueeze(0)
         spec_filename = filename.replace(".wav", ".spec.pt")
         if os.path.exists(spec_filename):

trainset_preprocess_pipeline_print.py

@@ -59,19 +59,34 @@ class PreProcess:
         wavfile.write(
             "%s/%s_%s.wav" % (self.gt_wavs_dir, idx0, idx1),
             self.sr,
-            (tmp_audio * 32768).astype(np.int16),
+            (tmp_audio * 1).astype(np.float32),
         )
-        tmp_audio = librosa.resample(tmp_audio, orig_sr=self.sr, target_sr=16000)
+
+        # default resample type of librosa.resample is "soxr_hq".
+        # Quality: soxr_vhq > soxr_hq
+        tmp_audio = librosa.resample(tmp_audio, orig_sr=self.sr, target_sr=16000, res_type="soxr_vhq")
+        tmp_audio = (tmp_audio / np.abs(tmp_audio).max() * (self.max * self.alpha)) + (
+            1 - self.alpha
+        ) * tmp_audio
+        wavfile.write(
+            "%s/%s_%s.wav" % (self.gt_wavs_dir, idx0, idx1),
+            self.sr,
+            (tmp_audio * 1).astype(np.float32),
+        )
+
         wavfile.write(
             "%s/%s_%s.wav" % (self.wavs16k_dir, idx0, idx1),
             16000,
-            (tmp_audio * 32768).astype(np.int16),
+            (tmp_audio * 1).astype(np.float32),
         )
 
     def pipeline(self, path, idx0):
         try:
             audio = load_audio(path, self.sr)
-            audio = signal.filtfilt(self.bh, self.ah, audio)
+            # zero phased digital filter cause pre-ringing noise...
+            # audio = signal.filtfilt(self.bh, self.ah, audio) 
+            audio = signal.lfilter(self.bh, self.ah, audio)
+
             idx1 = 0
             for audio in self.slicer.slice(audio):
                 i = 0