Update app.py

app.py

@@ -1,68 +1,98 @@
 import streamlit as st
+import requests
 import torchaudio
 from audiocraft.models import MusicGen
-import requests
 from pydub import AudioSegment
 import os
+import time
 
-# Function to generate voice with Resemble AI
-def generate_resemble_audio(api_key, project_uuid, text, voice_uuid):
-    url = f"https://app.resemble.ai/api/v2/projects/{project_uuid}/clips"
-    headers = {"Authorization": f"Token {Gpn4u6ObHlmLWc5aQvs8cQtt}"}
-    data = {"voice_uuid": voice_uuid, "text": text, "is_active": True}
-    response = requests.post(url, headers=headers, json=data)
-
-    if response.status_code == 200:
-        clip = response.json()
-        return clip["audio_src"]
-    else:
-        raise Exception(f"Failed to generate audio: {response.text}")
+# 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"
 
 # Streamlit UI
-st.set_page_config(page_title="Suno-like Music App with Resemble AI", layout="centered")
-st.title("Suno-like AI Music Generator with Resemble AI")
+st.set_page_config(page_title="AI Music Generator", layout="centered")
+st.title("AI Music Generator like Suno")
+
+# Step 1: User Lyrics Input
+st.subheader("Step 1: Enter Your Lyrics")
+lyrics = st.text_area("Lyrics (max 300 characters)", max_chars=300)
+
+# 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)", "")
 
-# Step 1: Input lyrics and music prompt
-st.subheader("Step 1: Input Lyrics and Music Prompt")
-lyrics = st.text_area("Enter lyrics:", "Life is good, the sky is blue, moving fast in brand new shoes")
-prompt = st.text_input("Enter music style:", "moody trap instrumental with 808 bass")
+# Function to call Resemble API
+def generate_voice_with_resemble(text, project_uuid):
+    headers = {
+        "Authorization": f"Token {RESEMBLE_API_TOKEN}",
+        "Accept": "application/json",
+        "Content-Type": "application/json"
+    }
+    data = {
+        "project_uuid": project_uuid,
+        "title": "streamlit_clip",
+        "body": text,
+        "voice": "your_custom_voice_id_if_needed"
+    }
+    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")
 
-# Step 2: Resemble AI credentials
-st.subheader("Step 2: Resemble AI Settings")
-api_key = st.text_input("Gpn4u6ObHlmLWc5aQvs8cQtt", type="password")
-project_id = st.text_input("Resemble Project UUID")
-voice_id = st.text_input("Resemble Voice UUID")
+    if not clip_id:
+        st.error("Failed to create clip on Resemble.")
+        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
+
+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")
+    else:
+        st.error("Voice generation failed.")
 
 # Step 3: Generate Instrumental
-output_path = "musicgen_output.wav"
-if st.button("Generate Instrumental"):
-    with st.spinner("Generating music with MusicGen..."):
+st.subheader("Step 3: Generate Instrumental")
+prompt = st.text_input("Describe the music (e.g. energetic trap beat)", "emotional trap beat")
+
+if st.button("Generate Music"):
+    with st.spinner("Generating music..."):
         model = MusicGen.get_pretrained("facebook/musicgen-small")
-        model.set_generation_params(duration=10)
+        model.set_generation_params(duration=5)
         output = model.generate([prompt])
-        torchaudio.save(output_path, output[0].cpu(), 32000)
-        st.audio(output_path, format="audio/wav")
-        st.success("Music generated successfully!")
+        torchaudio.save(MUSICGEN_OUTPUT_PATH, output[0].cpu(), 32000)
+        st.audio(MUSICGEN_OUTPUT_PATH, format="audio/wav")
+        st.success("Instrumental generated!")
 
-# Step 4: Generate Voice with Resemble and Combine
-if st.button("Generate Voice and Combine"):
+# 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:
-        # Generate voice using Resemble
-        audio_url = generate_resemble_audio(api_key, project_id, lyrics, voice_id)
-        voice_file = "resemble_voice.mp3"
-        with open(voice_file, "wb") as f:
-            f.write(requests.get(audio_url).content)
-
-        # Combine voice with music
-        beat = AudioSegment.from_wav(output_path)
-        voice = AudioSegment.from_mp3(voice_file)
+        beat = AudioSegment.from_wav(MUSICGEN_OUTPUT_PATH)
+        voice = AudioSegment.from_mp3(AUDIO_OUTPUT_PATH)
         voice = voice - 4
         beat = beat - 2
-        combined = beat.overlay(voice, loop=False)
-
-        combined_path = "final_mix.mp3"
-        combined.export(combined_path, format="mp3")
-        st.audio(combined_path, format="audio/mp3")
-        st.success("Final mix is ready!")
+        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(str(e))
+        st.error(f"Error combining files: {e}")
+else:
+    st.info("Generate both voice and music to combine.")