feat: start adding server code

generate.py

@@ -1,13 +1,10 @@
 import torchaudio
-import time
 from audiocraft.models.musicgen import MusicGen
 from audiocraft.data.audio import audio_write
 
 MODEL_NAME = "facebook/musicgen-large"
 MUSIC_DURATION_SECONDS = 60
 
-print(f"getting {MODEL_NAME}...")
-
 model = MusicGen.get_pretrained(MODEL_NAME)
 model.set_generation_params(duration=MUSIC_DURATION_SECONDS)
 descriptions = [
@@ -15,25 +12,20 @@ descriptions = [
     "calm, piano lo-fi beats to help with studying and focusing",
     "gentle lo-fi hip-hop to relax to",
     "gentle, quiet synthwave lo-fi beats",
-    "morning lo-fi beats"
+    "morning lo-fi beats",
 ]
 
-print("model obtained. generating wav files...")
 
-a = time.time()
+def generate(offset=0):
+    wav = model.generate(descriptions)
 
-wav = model.generate(descriptions)
-
-b = time.time()
-
-print(f"{len(wav)} generated. took {b - a} seconds.")
-
-for idx, one_wav in enumerate(wav):
-    # Will save under {idx}.wav, with loudness normalization at -14 db LUFS.
-    audio_write(
-        f"{idx}",
-        one_wav.cpu(),
-        model.sample_rate,
-        strategy="loudness",
-        loudness_compressor=True,
-    )
+    for idx, one_wav in enumerate(wav):
+        # Will save under {idx}.wav, with loudness normalization at -14 db LUFS.
+        audio_write(
+            f"{idx + offset}",
+            one_wav.cpu(),
+            model.sample_rate,
+            format="mp3",
+            strategy="loudness",
+            loudness_compressor=True,
+        )