From 49a3b0f2ae35d6c1b1851ba9f28006f52d5b0452 Mon Sep 17 00:00:00 2001
From: ylzz1997 <ylzz1997@outlook.com>
Date: Thu, 25 May 2023 01:58:24 +0800
Subject: [PATCH] Restore Loud match logic

---
 preprocess_hubert_f0.py |  4 ++--
 requirements.txt        |  3 +--
 requirements_win.txt    |  3 +--
 resample.py             | 21 +++++++--------------
 4 files changed, 11 insertions(+), 20 deletions(-)

diff --git a/preprocess_hubert_f0.py b/preprocess_hubert_f0.py
index b0f6822..9717f57 100644
--- a/preprocess_hubert_f0.py
+++ b/preprocess_hubert_f0.py
@@ -106,10 +106,10 @@ def process_one(filename, hmodel,f0p,diff=False,mel_extractor=None):
 
 
 def process_batch(filenames,f0p,diff=False,mel_extractor=None):
-    print("Loading hubert for content...")
+    print("Loading speech encoder for content...")
     device = "cuda" if torch.cuda.is_available() else "cpu"
     hmodel = utils.get_speech_encoder(speech_encoder,device=device)
-    print("Loaded hubert.")
+    print("Loaded speech encoder.")
     for filename in tqdm(filenames):
         process_one(filename, hmodel,f0p,diff,mel_extractor)
 
diff --git a/requirements.txt b/requirements.txt
index 99ddff5..7712dbd 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -20,5 +20,4 @@ tensorboard
 tensorboardX
 edge_tts
 pyyaml
-pynvml
-pyloudnorm
\ No newline at end of file
+pynvml
\ No newline at end of file
diff --git a/requirements_win.txt b/requirements_win.txt
index 155a3d3..f14759a 100644
--- a/requirements_win.txt
+++ b/requirements_win.txt
@@ -23,5 +23,4 @@ onnxoptimizer
 tensorboardX
 edge_tts
 pyyaml
-pynvml
-pyloudnorm
\ No newline at end of file
+pynvml
\ No newline at end of file
diff --git a/resample.py b/resample.py
index ca1732e..b28a86e 100644
--- a/resample.py
+++ b/resample.py
@@ -5,7 +5,6 @@ import numpy as np
 from multiprocessing import Pool, cpu_count
 from scipy.io import wavfile
 from tqdm import tqdm
-import pyloudnorm as pyln
 
 
 def process(item):
@@ -22,19 +21,13 @@ def process(item):
             wav = 0.98 * wav / peak
         wav2 = librosa.resample(wav, orig_sr=sr, target_sr=args.sr2)
         wav2 /= max(wav2.max(), -wav2.min())
-        try:
-            meter = pyln.Meter(args.sr2)
-            loudness = meter.integrated_loudness(wav2)
-            wav2 = pyln.normalize.loudness(wav2, loudness, -23.0)
-            # wav2 /= max(wav2.max(), -wav2.min())
-            save_name = wav_name
-            save_path2 = os.path.join(args.out_dir2, speaker, save_name)
-            wavfile.write(
-                save_path2,
-                args.sr2,
-                (wav2 * np.iinfo(np.int16).max).astype(np.int16))
-        except ValueError as e:
-            print(f"{wav_path} is too short(<400ms), the wav skip")
+        save_name = wav_name
+        save_path2 = os.path.join(args.out_dir2, speaker, save_name)
+        wavfile.write(
+            save_path2,
+            args.sr2,
+            (wav2 * np.iinfo(np.int16).max).astype(np.int16)
+        )