test

app.py

@@ -51,7 +51,7 @@ def infer_music(lrc, ref_audio_path, steps, max_frames=2048, device='cuda'):
 
 def R1_infer1(theme, tags_gen, language):
     try:
-        client = OpenAI(api_key=os.getenv('DP_API'), base_url="https://api.deepseek.com")
+        client = OpenAI(api_key=os.getenv('HS_DP_API'), base_url = "https://ark.cn-beijing.volces.com/api/v3")
 
         llm_prompt = """
         请围绕"{theme}"主题生成一首符合"{tags}"风格的完整歌词。生成的{language}语言的歌词。
@@ -66,7 +66,7 @@ def R1_infer1(theme, tags_gen, language):
         """
 
         response = client.chat.completions.create(
-            model='deepseek-reasoner',
+            model="ep-20250304144033-nr9wl",
             messages=[
                 {"role": "system", "content": "You are a professional musician who has been invited to make music-related comments."},
                 {"role": "user", "content": llm_prompt.format(theme=theme, tags=tags_gen, language=language)},
@@ -85,14 +85,14 @@ def R1_infer1(theme, tags_gen, language):
 
 
 def R1_infer2(tags_lyrics, lyrics_input):
-    client = OpenAI(api_key=os.getenv('DP_API'), base_url="https://api.deepseek.com")
+    client = OpenAI(api_key=os.getenv('HS_DP_API'), base_url = "https://ark.cn-beijing.volces.com/api/v3")
 
     llm_prompt = """
     {lyrics_input}这是一首歌的歌词,每一行是一句歌词,{tags_lyrics}是我希望这首歌的风格，我现在想要给这首歌的每一句歌词打时间戳得到LRC，我希望时间戳分配应根据歌曲的标签、歌词的情感、节奏来合理推测，而非机械地按照歌词长度分配。第一句歌词的时间戳应考虑前奏长度，避免歌词从 `[00:00.00]` 直接开始。严格按照 LRC 格式输出歌词，每行格式为 `[mm:ss.xx]歌词内容`。最后的结果只输出LRC,不需要其他的解释。
     """
 
     response = client.chat.completions.create(
-        model='deepseek-reasoner',
+        model="ep-20250304144033-nr9wl",
         messages=[
             {"role": "system", "content": "You are a professional musician who has been invited to make music-related comments."},
             {"role": "user", "content": llm_prompt.format(lyrics_input=lyrics_input, tags_lyrics=tags_lyrics)},
@@ -128,7 +128,7 @@ with gr.Blocks(css=css) as demo:
     gr.Markdown("<h1 style='text-align: center'>DiffRhythm (谛韵)</h1>")
     gr.HTML("""
         <div style="display:flex; justify-content: center; column-gap:4px;">
-            <a href="https://github.com/ASLP-lab/DiffRhythm">
+            <a href="https://arxiv.org/abs/2503.01183">
                 <img src='https://img.shields.io/badge/Arxiv-Paper-blue'>
             </a> 
             <a href="https://github.com/ASLP-lab/DiffRhythm">

requirements.txt

@@ -30,5 +30,4 @@ einops==0.8.1
 lazy_loader==0.4
 scipy==1.15.2
 ftfy==6.3.1
-torchdiffeq==0.2.5
-https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl
+torchdiffeq==0.2.5