Update app.py

app.py

@@ -62,14 +62,13 @@ voice_presets_dict = {
     format_speaker_key(key): key for key in voice_presets
 }
 
-model = BarkModel.from_pretrained("suno/bark", torch_dtype=torch.float16).to(device)
+model = BarkModel.from_pretrained("suno/bark", torch_dtype=torch.float16, use_flash_attention_2=True).to(device)
 sampling_rate = model.generation_config.sample_rate
 silence = np.zeros(int(0.25 * sampling_rate))  # quarter second of silence
 voice_preset = "v2/en_speaker_6"
 
-# convert to bettertransformer
-model = model.to_bettertransformer()
-BATCH_SIZE = 12
+model = model.enable_cpu_offload()
+BATCH_SIZE = 32
 
 # enable CPU offload
 model.enable_cpu_offload()