2 pure en

app.py

@@ -8,24 +8,25 @@ import numpy as np
 import gradio as gr
 import librosa.display
 import matplotlib.pyplot as plt
-from utils import get_modelist, find_wav_files, embed_img, TEMP_DIR
 from model import EvalNet
+from utils import get_modelist, find_wav_files, embed_img
 
 
 TRANSLATE = {
-    "vibrato": "揉弦 Rou xian",
-    "trill": "颤音 Chan yin",
-    "tremolo": "颤弓 Chan gong",
-    "staccato": "顿弓 Dun gong",
-    "ricochet": "抛弓 Pao gong",
-    "pizzicato": "拨弦 Bo xian",
-    "percussive": "击弓 Ji gong",
-    "legato_slide_glissando": "连滑音 Lian hua yin",
-    "harmonic": "泛音 Fan yin",
-    "diangong": "垫弓 Dian gong",
-    "detache": "分弓 Fen gong",
+    "vibrato": "Rou xian",
+    "trill": "Chan yin",
+    "tremolo": "Chan gong",
+    "staccato": "Dun gong",
+    "ricochet": "Pao gong",
+    "pizzicato": "Bo xian",
+    "percussive": "Ji gong",
+    "legato_slide_glissando": "Lian hua yin",
+    "harmonic": "Fan yin",
+    "diangong": "Dian gong",
+    "detache": "Fen gong",
 }
 CLASSES = list(TRANSLATE.keys())
+TEMP_DIR = "./__pycache__/tmp"
 SAMPLE_RATE = 44100
 
 
@@ -107,7 +108,7 @@ def infer(wav_path: str, log_name: str, folder_path=TEMP_DIR):
         shutil.rmtree(folder_path)
 
     if not wav_path:
-        return None, "请输入音频 Please input an audio!"
+        return None, "Please input an audio!"
 
     try:
         model = EvalNet(log_name, len(TRANSLATE)).model
@@ -138,30 +139,26 @@ if __name__ == "__main__":
         gr.Interface(
             fn=infer,
             inputs=[
-                gr.Audio(label="上传录音 Upload a recording", type="filepath"),
-                gr.Dropdown(
-                    choices=models, label="选择模型 Select a model", value=models[0]
-                ),
+                gr.Audio(label="Upload a recording", type="filepath"),
+                gr.Dropdown(choices=models, label="Select a model", value=models[0]),
             ],
             outputs=[
-                gr.Textbox(label="音频文件名 Audio filename", show_copy_button=True),
-                gr.Textbox(
-                    label="演奏技法识别 Playing tech recognition", show_copy_button=True
-                ),
+                gr.Textbox(label="Audio filename", show_copy_button=True),
+                gr.Textbox(label="Playing tech recognition", show_copy_button=True),
             ],
             examples=examples,
             cache_examples=False,
             allow_flagging="never",
-            title="建议录音时长保持在 3s 左右<br>It is recommended to keep the recording length around 3s.",
+            title="It is recommended to keep the recording length around 3s.",
         )
 
         gr.Markdown(
             """
-# 引用 Cite
+# Cite
 ```bibtex
 @dataset{zhaorui_liu_2021_5676893,
   author       = {Monan Zhou, Shenyang Xu, Zhaorui Liu, Zhaowen Wang, Feng Yu, Wei Li and Baoqiang Han},
-  title        = {CCMusic: an Open and Diverse Database for Chinese and General Music Information Retrieval Research},
+  title        = {CCMusic: an Open and Diverse Database for Chinese Music Information Retrieval Research},
   month        = {mar},
   year         = {2024},
   publisher    = {HuggingFace},

requirements.txt

@@ -1,6 +1,6 @@
-librosa
 torch
+pillow
+librosa
 matplotlib
 torchvision
-pillow
-modelscope==1.15
+modelscope[framework]==1.18

utils.py

@@ -5,10 +5,9 @@ from modelscope import snapshot_download
 from PIL import Image
 
 MODEL_DIR = snapshot_download(
-    f"ccmusic-database/erhu_playing_tech",
-    cache_dir=f"{os.getcwd()}/__pycache__",
+    "ccmusic-database/erhu_playing_tech",
+    cache_dir="./__pycache__",
 )
-TEMP_DIR = f"{os.getcwd()}/flagged"
 
 
 def toCUDA(x):
@@ -34,22 +33,17 @@ def get_modelist(model_dir=MODEL_DIR):
     try:
         entries = os.listdir(model_dir)
     except OSError as e:
-        print(f"无法访问 {model_dir}: {e}")
+        print(f"Cannot access {model_dir}: {e}")
         return
 
-    # 遍历所有条目
     output = []
     for entry in entries:
-        # 获取完整路径
         full_path = os.path.join(model_dir, entry)
-        # 跳过'.git'文件夹
         if entry == ".git" or entry == "examples":
-            print(f"跳过 .git 或 examples 文件夹: {full_path}")
+            print(f"Skip .git / examples dir: {full_path}")
             continue
 
-        # 检查条目是文件还是目录
         if os.path.isdir(full_path):
-            # 打印目录路径
             output.append(os.path.basename(full_path))
 
     return output