From ade5d641e83c960afa78ad50429bf391d5b17217 Mon Sep 17 00:00:00 2001
From: litagin02 <litagin02@gmail.com>
Date: Mon, 8 Jan 2024 16:21:36 +0900
Subject: [PATCH 1/2] Fix bug: add missing get_style_vector function

---
 style_gen.py | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/style_gen.py b/style_gen.py
index 7f74243..12dab20 100644
--- a/style_gen.py
+++ b/style_gen.py
@@ -7,8 +7,9 @@ import torch
 from tqdm import tqdm
 
 import utils
-from config import config
+from common.log import logger
 from common.stdout_wrapper import SAFE_STDOUT
+from config import config
 
 warnings.filterwarnings("ignore", category=UserWarning)
 from pyannote.audio import Inference, Model
@@ -19,12 +20,16 @@ device = torch.device(config.style_gen_config.device)
 inference.to(device)
 
 
+# 推論時にインポートするために短いが関数を書く
+def get_style_vector(wav_path):
+    return inference(wav_path)
+
+
 def save_style_vector(wav_path):
     try:
-        style_vec = inference(wav_path)
+        style_vec = get_style_vector(wav_path)
     except Exception as e:
-        print(f"\nError occurred with file: {wav_path}")
-        print(e)
+        logger.error(f"\nError occurred with file: {wav_path}, Details:\n{e}\n")
         raise
     np.save(f"{wav_path}.npy", style_vec)  # `test.wav` -> `test.wav.npy`
     return style_vec
@@ -69,4 +74,4 @@ if __name__ == "__main__":
             )
         )
 
-    print(f"Finished generating style vectors! total: {len(wavnames)} npy files.")
+    logger.info(f"Finished generating style vectors! total: {len(wavnames)} npy files.")

From fb43a178a38b88ec6942ba97770eead3a8de6955 Mon Sep 17 00:00:00 2001
From: litagin02 <litagin02@gmail.com>
Date: Mon, 8 Jan 2024 16:22:32 +0900
Subject: [PATCH 2/2] Fix bug: add missing get_style_vector function

---
 common/tts_model.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/common/tts_model.py b/common/tts_model.py
index f83a806..307c6c3 100644
--- a/common/tts_model.py
+++ b/common/tts_model.py
@@ -61,9 +61,9 @@ class Model:
     def get_style_vector_from_audio(
         self, audio_path: str, weight: float = 1.0
     ) -> np.ndarray:
-        from style_gen import extract_style_vector
+        from style_gen import get_style_vector
 
-        xvec = extract_style_vector(audio_path)
+        xvec = get_style_vector(audio_path)
         mean = self.style_vectors[0]
         xvec = mean + (xvec - mean) * weight
         return xvec