Simplify music generation interface by removing audio prompts

- Removed audio prompt tab and related functionality
- Switched to text-only prompts with default value
- Updated Best Practices guide to reflect text prompt format
- Fixed typo in "coming soon" text

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

app.py

@@ -31,7 +31,7 @@ cfm, tokenizer, muq, vae = prepare_model(device)
 cfm = torch.compile(cfm)
 
 @spaces.GPU(duration=20)
-def infer_music(lrc, ref_audio_path, text_prompt, current_prompt_type, seed=42, randomize_seed=False, steps=32, cfg_strength=4.0, file_type='wav', odeint_method='euler', max_frames=2048, device='cuda'):
+def infer_music(lrc, text_prompt, seed=42, randomize_seed=False, steps=32, cfg_strength=4.0, file_type='wav', odeint_method='euler', max_frames=2048, device='cuda'):
 
     if randomize_seed:
         seed = random.randint(0, MAX_SEED)
@@ -40,10 +40,7 @@ def infer_music(lrc, ref_audio_path, text_prompt, current_prompt_type, seed=42,
     vocal_flag = False
     try:
         lrc_prompt, start_time = get_lrc_token(lrc, tokenizer, device)
-        if current_prompt_type == 'audio':
-            style_prompt, vocal_flag = get_audio_style_prompt(muq, ref_audio_path)
-        else:
-            style_prompt = get_text_style_prompt(muq, text_prompt)
+        style_prompt = get_text_style_prompt(muq, text_prompt)
     except Exception as e:
         raise gr.Error(f"Error: {str(e)}")
     negative_style_prompt = get_negative_style_prompt(device)
@@ -188,22 +185,11 @@ with gr.Blocks(css=css) as demo:
                         value="""[00:10.00]Moonlight spills through broken blinds\n[00:13.20]Your shadow dances on the dashboard shrine\n[00:16.85]Neon ghosts in gasoline rain\n[00:20.40]I hear your laughter down the midnight train\n[00:24.15]Static whispers through frayed wires\n[00:27.65]Guitar strings hum our cathedral choirs\n[00:31.30]Flicker screens show reruns of June\n[00:34.90]I'm drowning in this mercury lagoon\n[00:38.55]Electric veins pulse through concrete skies\n[00:42.10]Your name echoes in the hollow where my heartbeat lies\n[00:45.75]We're satellites trapped in parallel light\n[00:49.25]Burning through the atmosphere of endless night\n[01:00.00]Dusty vinyl spins reverse\n[01:03.45]Our polaroid timeline bleeds through the verse\n[01:07.10]Telescope aimed at dead stars\n[01:10.65]Still tracing constellations through prison bars\n[01:14.30]Electric veins pulse through concrete skies\n[01:17.85]Your name echoes in the hollow where my heartbeat lies\n[01:21.50]We're satellites trapped in parallel light\n[01:25.05]Burning through the atmosphere of endless night\n[02:10.00]Clockwork gears grind moonbeams to rust\n[02:13.50]Our fingerprint smudged by interstellar dust\n[02:17.15]Velvet thunder rolls through my veins\n[02:20.70]Chasing phantom trains through solar plane\n[02:24.35]Electric veins pulse through concrete skies\n[02:27.90]Your name echoes in the hollow where my heartbeat lies"""    
                     )
                     
-                    current_prompt_type = gr.State(value="audio")
-                    with gr.Tabs() as inside_tabs:
-                        with gr.Tab("Audio Prompt"):
-                            audio_prompt = gr.Audio(label="Audio Prompt", type="filepath", value="./src/prompt/default.wav")
-                        with gr.Tab("Text Prompt"):
-                            text_prompt = gr.Textbox(
-                            label="Text Prompt",
-                            placeholder="Enter the Text Prompt, eg: emotional piano pop",
-                            )
-                        def update_prompt_type(evt: gr.SelectData):
-                            return "audio" if evt.index == 0 else "text"
-
-                        inside_tabs.select(
-                            fn=update_prompt_type,
-                            outputs=current_prompt_type
-                        )
+                    text_prompt = gr.Textbox(
+                        label="Text Prompt",
+                        placeholder="Enter the Text Prompt, eg: emotional piano pop",
+                        value="Pop Emotional Piano"
+                    )
                     
                 with gr.Column():
                     with gr.Accordion("Best Practices Guide", open=True):
@@ -218,13 +204,13 @@ with gr.Blocks(css=css) as demo:
 2. **Generation Duration Limits**
     - Current version supports maximum **95 seconds** of music generation
     - Total timestamps should not exceed 01:35.00 (95 seconds)
-3. **Audio Prompt Requirements**
-    - Reference audio should be ≥ 1 second, audio >10 seconds will be randomly clipped into 10 seconds
-    - For optimal results, the 10-second clips should be carefully selected
-    - Shorter clips may lead to incoherent generation
+3. **Text Prompt Format**
+    - Use descriptive terms for style like "pop", "rock", "jazz"
+    - Add emotions like "emotional", "upbeat", "melancholic"
+    - Include instruments like "piano", "guitar", "orchestral"
 4. **Supported Languages**
     - **Chinese and English**
-    - More languages comming soon
+    - More languages coming soon
 
 5. **Others** 
     - If loading audio result is slow, you can select Output Format as mp3 in Advanced Settings.                                
@@ -265,27 +251,6 @@ with gr.Blocks(css=css) as demo:
                         file_type = gr.Dropdown(["wav", "mp3", "ogg"], label="Output Format", value="wav")
 
 
-            gr.Examples(
-                examples=[
-                    ["./src/prompt/pop_cn.wav"], 
-                    ["./src/prompt/pop_en.wav"], 
-                    ["./src/prompt/rock_cn.wav"], 
-                    ["./src/prompt/rock_en.wav"], 
-                    ["./src/prompt/country_cn.wav"], 
-                    ["./src/prompt/country_en.wav"],
-                    ["./src/prompt/classic_cn.wav"],
-                    ["./src/prompt/classic_en.wav"],
-                    ["./src/prompt/jazz_cn.wav"],
-                    ["./src/prompt/jazz_en.wav"],
-                    ["./src/prompt/rap_cn.wav"],
-                    ["./src/prompt/rap_en.wav"],
-                    ["./src/prompt/default.wav"]
-                ],
-                inputs=[audio_prompt],  
-                label="Audio Examples",
-                examples_per_page=13,
-                elem_id="audio-examples-container" 
-            )
             
             gr.Examples(
                 examples=[
@@ -403,7 +368,7 @@ with gr.Blocks(css=css) as demo:
     
     lyrics_btn.click(
         fn=infer_music,
-        inputs=[lrc, audio_prompt, text_prompt, current_prompt_type, seed, randomize_seed, steps, cfg_strength, file_type, odeint_method],
+        inputs=[lrc, text_prompt, seed, randomize_seed, steps, cfg_strength, file_type, odeint_method],
         outputs=audio_output
     )