Update app.py

app.py

@@ -1,98 +1,101 @@
 import streamlit as st
 import requests
+import os
 import torchaudio
-from audiocraft.models import MusicGen
 from pydub import AudioSegment
-import os
-import time
+from audiocraft.models import MusicGen
+from tempfile import NamedTemporaryFile
 
-# Constants
-AUDIO_OUTPUT_PATH = "voice_output.mp3"
-COMBINED_OUTPUT_PATH = "combined_output.mp3"
-MUSICGEN_OUTPUT_PATH = "musicgen_output.wav"
-RESEMBLE_API_TOKEN = "your_resemble_api_token_here"
+# CONFIGURATION
+RESEMBLE_API_KEY = os.getenv("RESEMBLE_API_KEY", "your_resemble_api_key")
+PROJECT_UUID = "96e7c9b4-voice"
 
-# Streamlit UI
-st.set_page_config(page_title="AI Music Generator", layout="centered")
-st.title("AI Music Generator like Suno")
+st.set_page_config(page_title="Suno-Like Music App", layout="centered")
+st.title("Suno-Like AI Music Generator")
 
-# Step 1: User Lyrics Input
-st.subheader("Step 1: Enter Your Lyrics")
-lyrics = st.text_area("Lyrics (max 300 characters)", max_chars=300)
+# Lyrics Input
+st.subheader("Step 1: Enter Lyrics")
+lyrics = st.text_area("Write your lyrics here:", "We go hard, never stop, always chasing dreams")
 
-# Step 2: Generate Voice with Resemble AI
-st.subheader("Step 2: Generate Voice")
-project_uuid = st.text_input("Enter your Resemble Project UUID", "96e7c9b4-voice")
-voice_clip_uuid = st.text_input("Enter your Voice Clip UUID (or leave blank to generate new)", "")
+# Prompt Input
+st.subheader("Step 2: Describe Your Beat")
+prompt = st.text_input("Beat style (e.g. dark trap, chill pop):", "dark emotional trap beat with heavy bass")
+
+# Generate Instrumental
+music_path = None
+if st.button("Generate Music"):
+    with st.spinner("Generating instrumental..."):
+        model = MusicGen.get_pretrained("facebook/musicgen-small")
+        model.set_generation_params(duration=10)
+        output = model.generate([prompt])
+        music_path = "output_music.wav"
+        torchaudio.save(music_path, output[0].cpu(), 32000)
+        st.audio(music_path, format="audio/wav")
+        st.success("Instrumental ready!")
 
-# Function to call Resemble API
-def generate_voice_with_resemble(text, project_uuid):
+# Generate Voice with Resemble AI
+def generate_voice_with_resemble(lyrics, project_uuid):
+    url = f"https://app.resemble.ai/api/v2/projects/{project_uuid}/clips"
     headers = {
-        "Authorization": f"Token {RESEMBLE_API_TOKEN}",
-        "Accept": "application/json",
+        "Authorization": f"Token {RESEMBLE_API_KEY}",
         "Content-Type": "application/json"
     }
-    data = {
-        "project_uuid": project_uuid,
-        "title": "streamlit_clip",
-        "body": text,
-        "voice": "your_custom_voice_id_if_needed"
+
+    payload = {
+        "title": "AI Generated Clip",
+        "body": lyrics,
+        "is_active": True
     }
-    res = requests.post("https://app.resemble.ai/api/v2/clips", headers=headers, json=data)
-    res_data = res.json()
-    clip_id = res_data.get("item", {}).get("uuid")
 
-    if not clip_id:
-        st.error("Failed to create clip on Resemble.")
+    response = requests.post(url, headers=headers, json=payload)
+
+    if response.status_code != 200:
+        st.error(f"Resemble API Error: {response.status_code}")
+        st.text(response.text)
         return None
 
-    # Polling for completion
-    for _ in range(30):
-        clip_res = requests.get(f"https://app.resemble.ai/api/v2/clips/{clip_id}", headers=headers)
-        clip_info = clip_res.json()
-        if clip_info.get("is_complete"):
-            audio_url = clip_info.get("audio_src")
-            audio_data = requests.get(audio_url).content
-            with open(AUDIO_OUTPUT_PATH, "wb") as f:
-                f.write(audio_data)
-            return AUDIO_OUTPUT_PATH
-        time.sleep(3)
-    return None
+    res_data = response.json()
+    clip_url = res_data.get("item", {}).get("audio_src")
+    return clip_url
 
-if st.button("Generate Voice") and lyrics:
-    st.info("Requesting Resemble to generate voice...")
-    path = generate_voice_with_resemble(lyrics, project_uuid)
-    if path and os.path.exists(path):
-        st.audio(path, format="audio/mp3")
+voice_path = None
+if st.button("Generate Voice with Resemble"):
+    if not lyrics:
+        st.warning("Please enter lyrics first.")
     else:
-        st.error("Voice generation failed.")
+        with st.spinner("Generating voice..."):
+            audio_url = generate_voice_with_resemble(lyrics, PROJECT_UUID)
+            if audio_url:
+                response = requests.get(audio_url)
+                if response.status_code == 200:
+                    with NamedTemporaryFile(delete=False, suffix=".mp3") as f:
+                        f.write(response.content)
+                        voice_path = f.name
+                    st.audio(voice_path, format="audio/mp3")
+                    st.success("Voice generation complete!")
+                else:
+                    st.error("Failed to download audio from Resemble.")
+            else:
+                st.error("Voice generation failed.")
 
-# Step 3: Generate Instrumental
-st.subheader("Step 3: Generate Instrumental")
-prompt = st.text_input("Describe the music (e.g. energetic trap beat)", "emotional trap beat")
+# Combine Voice and Instrumental
+if st.button("Combine Voice and Beat"):
+    if not music_path or not voice_path:
+        st.warning("Generate both voice and music before combining.")
+    else:
+        try:
+            beat = AudioSegment.from_wav(music_path)
+            voice = AudioSegment.from_mp3(voice_path)
 
-if st.button("Generate Music"):
-    with st.spinner("Generating music..."):
-        model = MusicGen.get_pretrained("facebook/musicgen-small")
-        model.set_generation_params(duration=5)
-        output = model.generate([prompt])
-        torchaudio.save(MUSICGEN_OUTPUT_PATH, output[0].cpu(), 32000)
-        st.audio(MUSICGEN_OUTPUT_PATH, format="audio/wav")
-        st.success("Instrumental generated!")
+            voice = voice - 2
+            beat = beat - 4
+
+            combined = beat.overlay(voice)
+            combined_path = "combined_output.mp3"
+            combined.export(combined_path, format="mp3")
 
-# Step 4: Combine
-st.subheader("Step 4: Combine Voice and Beat")
-if os.path.exists(MUSICGEN_OUTPUT_PATH) and os.path.exists(AUDIO_OUTPUT_PATH):
-    try:
-        beat = AudioSegment.from_wav(MUSICGEN_OUTPUT_PATH)
-        voice = AudioSegment.from_mp3(AUDIO_OUTPUT_PATH)
-        voice = voice - 4
-        beat = beat - 2
-        combined = beat.overlay(voice)
-        combined.export(COMBINED_OUTPUT_PATH, format="mp3")
-        st.audio(COMBINED_OUTPUT_PATH, format="audio/mp3")
-        st.success("Combined track ready!")
-    except Exception as e:
-        st.error(f"Error combining files: {e}")
-else:
-    st.info("Generate both voice and music to combine.")
+            st.audio(combined_path, format="audio/mp3")
+            st.success("Final output ready!")
+        except Exception as e:
+            st.error(f"Error while c
+            ombining: {e}")