From b664bc77771d3a84d931a57d0cdd701ca925860a Mon Sep 17 00:00:00 2001
From: Yushen CHEN <45333109+SWivid@users.noreply.github.com>
Date: Fri, 1 Nov 2024 18:20:39 +0800
Subject: [PATCH] Update finetune_gradio.py

---
 src/f5_tts/train/finetune_gradio.py | 12 ++----------
 1 file changed, 2 insertions(+), 10 deletions(-)

diff --git a/src/f5_tts/train/finetune_gradio.py b/src/f5_tts/train/finetune_gradio.py
index beb509a83..9b46b0102 100644
--- a/src/f5_tts/train/finetune_gradio.py
+++ b/src/f5_tts/train/finetune_gradio.py
@@ -28,7 +28,7 @@
 from scipy.io import wavfile
 from transformers import pipeline
 from cached_path import cached_path
-from f5_tts.api import F5TTS, target_sample_rate
+from f5_tts.api import F5TTS
 from f5_tts.model.utils import convert_char_to_pinyin
 from importlib.resources import files
 
@@ -174,15 +174,7 @@ def load_settings(project_name):
 def get_audio_duration(audio_path):
     """Calculate the duration mono of an audio file."""
     audio, sample_rate = torchaudio.load(audio_path)
-
-    if audio.shape[0] > 1:
-        audio = torch.mean(audio, dim=0, keepdim=True)
-
-    if sample_rate != target_sample_rate:
-        audio = torchaudio.transforms.Resample(sample_rate, target_sample_rate)
-
-    num_channels = audio.shape[0]
-    return audio.shape[1] / (sample_rate * num_channels)
+    return audio.shape[1] / sample_rate
 
 
 def clear_text(text):