Upload 2 files

app.py

@@ -1,16 +1,63 @@
-import gradio as gr
-from musicgen_utils import generate_music_prompt
-
-with gr.Blocks(title="🎼 AI Music & Tone Generator") as demo:
-    gr.Markdown("""
-    # 🎵 AI Tone & Music Generator
-    *Generated using real instrument‑like audio from a lightweight model*
-    """)
-    with gr.Tab("🎧 Music by Prompt"):
-        prompt = gr.Textbox(label="Describe your sound", placeholder="e.g. flute melody, bird chirping in forest, drum beat")
-        mood = gr.Radio(["Happy", "Calm", "Energetic", "Relaxed"], value="Calm", label="Mood")
-        tempo = gr.Slider(30, 200, value=120, step=5, label="Tempo (BPM)")
-        btn = gr.Button("Generate Audio 🎶")
-        audio_out = gr.Audio(label="Generated Sound", type="filepath")
-        btn.click(fn=generate_music_prompt, inputs=[prompt, mood, tempo], outputs=audio_out)
+
+    import gradio as gr
+    from musicgen_utils import (
+        generate_music_prompt,
+        remix_music,
+        generate_random,
+        category_to_text,
+        generate_tone,
+        export_midi_file
+    )
+
+    categories = [
+        "Bird Chirping", "Flute Solo", "Guitar Riff", "Drum Loop",
+        "Harmonium Chant", "Piano Arpeggio", "Violin Ambient",
+        "Bass Groove", "Ambient Forest", "Temple Bells"
+    ]
+
+    with gr.Blocks(title="🎼 AI Music & Tone Generator") as demo:
+        gr.Markdown("# 🎵 AI Music Generator
+_Easy, beginner-friendly audio & tone creation_")
+
+        with gr.Tabs():
+            with gr.Tab("🎧 Music from Prompt"):
+                with gr.Row():
+                    category = gr.Dropdown(choices=categories, label="🎼 Choose Instrument Category")
+                    prompt = gr.Textbox(lines=2, placeholder="Describe your music...", label="Prompt")
+                with gr.Row():
+                    mood = gr.Radio(["Happy", "Sad", "Calm", "Epic"], label="Mood", value="Calm")
+                    tempo = gr.Slider(60, 180, step=5, value=100, label="Tempo (BPM)")
+                    duration = gr.Slider(5, 30, step=1, value=10, label="Duration (seconds)")
+                    remix_strength = gr.Slider(0.5, 1.5, step=0.1, value=1.0, label="Remix Intensity")
+
+                with gr.Row():
+                    btn_gen = gr.Button("🎶 Generate Music")
+                    btn_remix = gr.Button("🪄 AI Remix")
+                    btn_rand = gr.Button("🎲 Randomize")
+
+                output_audio = gr.Audio(label="Output Music")
+                category.change(fn=category_to_text, inputs=category, outputs=prompt)
+                btn_gen.click(fn=generate_music_prompt, inputs=[prompt, mood, tempo, duration, remix_strength], outputs=output_audio)
+                btn_remix.click(fn=remix_music, inputs=[prompt, mood, tempo], outputs=output_audio)
+                btn_rand.click(fn=generate_random, outputs=output_audio)
+
+            with gr.Tab("🎹 Tone Synth (Fallback)"):
+                with gr.Row():
+                    wave = gr.Radio(["Sine", "Square", "Saw", "Triangle"], label="Waveform", value="Sine")
+                    freq = gr.Slider(100, 2000, value=440, step=10, label="Frequency (Hz)")
+                    duration2 = gr.Slider(1, 10, value=3, step=1, label="Duration (sec)")
+                    volume = gr.Slider(0.1, 1.0, value=0.5, step=0.1, label="Volume")
+
+                with gr.Row():
+                    bass = gr.Slider(-10, 10, value=0, label="Bass")
+                    mid = gr.Slider(-10, 10, value=0, label="Mid")
+                    treble = gr.Slider(-10, 10, value=0, label="Treble")
+
+                btn_tone = gr.Button("🎼 Generate Tone")
+                audio_out = gr.Audio(label="Tone Output")
+                midi_out = gr.File(label="Download MIDI")
+
+                btn_tone.click(fn=generate_tone, inputs=[wave, freq, duration2, volume, bass, mid, treble], outputs=audio_out)
+                btn_tone.click(fn=export_midi_file, inputs=[freq, duration2], outputs=midi_out)
+
     demo.launch()

musicgen_utils.py

@@ -1,17 +1,97 @@
+
 import torch
+import numpy as np
 from transformers import pipeline
 import tempfile
-import numpy as np
 from scipy.io import wavfile
+import random
+import math
+
+category_to_prompt = {
+    "Bird Chirping": "a forest with birds chirping",
+    "Flute Solo": "a calm flute solo in a forest",
+    "Guitar Riff": "a funky electric guitar riff",
+    "Drum Loop": "a groovy drum loop",
+    "Harmonium Chant": "indian harmonium prayer music",
+    "Piano Arpeggio": "beautiful piano arpeggios",
+    "Violin Ambient": "slow ambient violin",
+    "Bass Groove": "funky electric bass groove",
+    "Ambient Forest": "peaceful nature forest ambient",
+    "Temple Bells": "indian temple bells ringing"
+}
 
-# load once
-synth = pipeline("text-to-audio", model="facebook/musicgen-small", torch_dtype=torch.float32)
+# Try loading model
+try:
+    musicgen = pipeline("text-to-audio", model="facebook/musicgen-small", torch_dtype=torch.float32)
+    model_available = True
+except:
+    musicgen = None
+    model_available = False
 
-def generate_music_prompt(prompt, mood, tempo):
-    text = f"{mood} {prompt} at {tempo} bpm"
-    out = synth(text, forward_params={"do_sample": True, "max_new_tokens": 500})
-    rate = out["sampling_rate"]
-    data = np.array(out["audio"])
+def generate_music_prompt(prompt, mood, tempo, duration, remix_strength):
+    if not model_available:
+        return None
+    full_prompt = f"{mood} {prompt} at {tempo} bpm"
+    audio = musicgen(full_prompt, forward_params={
+        "do_sample": True,
+        "top_k": 250,
+        "temperature": remix_strength,
+        "max_new_tokens": int(duration * 50)
+    })
+    sr = audio["sampling_rate"]
+    data = np.array(audio["audio"])
     tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
-    wavfile.write(tmp.name, rate, data)
+    wavfile.write(tmp.name, sr, data)
     return tmp.name
+
+def remix_music(prompt, mood, tempo):
+    return generate_music_prompt(prompt, mood, tempo, duration=10, remix_strength=random.uniform(0.7, 1.0))
+
+def generate_random():
+    category = random.choice(list(category_to_prompt.keys()))
+    prompt = category_to_prompt[category]
+    mood = random.choice(["Happy", "Sad", "Calm", "Epic"])
+    return generate_music_prompt(prompt, mood, 100, 10, 1.0)
+
+def category_to_text(category):
+    return category_to_prompt.get(category, "")
+
+# Fallback waveform tone generation
+def generate_tone(waveform, freq, duration, volume, bass, mid, treble):
+    rate = 44100
+    t = np.linspace(0, duration, int(rate * duration), endpoint=False)
+    if waveform == "Sine":
+        wave = np.sin(2 * np.pi * freq * t)
+    elif waveform == "Square":
+        wave = np.sign(np.sin(2 * np.pi * freq * t))
+    elif waveform == "Saw":
+        wave = 2 * (t * freq - np.floor(0.5 + t * freq))
+    elif waveform == "Triangle":
+        wave = 2 * np.abs(2 * (t * freq - np.floor(t * freq + 0.5))) - 1
+    else:
+        wave = np.zeros_like(t)
+
+    # Apply EQ (simple gain)
+    wave = wave * volume
+    eq_gain = (1 + bass/10 + mid/10 + treble/10) / 3
+    wave *= eq_gain
+
+    wave = np.int16(wave * 32767)
+    tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
+    wavfile.write(tmp.name, rate, wave)
+    return tmp.name
+
+def export_midi_file(freq, duration):
+    try:
+        import mido
+        mid = mido.MidiFile()
+        track = mido.MidiTrack()
+        mid.tracks.append(track)
+        note = int(69 + 12 * math.log2(freq / 440))
+        track.append(mido.Message("note_on", note=note, velocity=64, time=0))
+        track.append(mido.Message("note_off", note=note, velocity=64, time=int(duration * 480)))
+        tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".mid")
+        mid.save(tmp.name)
+        return tmp.name
+    except:
+        return None