live api

live_api.py

@@ -6,7 +6,6 @@ import audresample
 import text_utils
 
 import re
-import srt
 import subprocess
 import markdown
 import json
@@ -14,10 +13,9 @@ from pathlib import Path
 from types import SimpleNamespace
 from flask import Flask, request, send_from_directory
 from flask_cors import CORS
-from audiocraft.audiogen import AudioGen #, audio_write
+from audiocraft.builders import AudioGen #, audio_write
 
-sound_generator = AudioGen.get_pretrained('facebook/audiogen-medium')
-sound_generator.set_generation_params(duration=.7)
+sound_generator = AudioGen(duration=.74, device='cuda:0').to('cuda:0').eval()
 
     
 # ====STYLE VECTOR====
@@ -48,7 +46,7 @@ Path(CACHE_DIR).mkdir(parents=True, exist_ok=True)
 def tts_multi_sentence(scene=None):
     if scene is not None and len(scene) >= 4:
         print(f'Processing: {scene} ..')
-        x = sound_generator.generate([scene])[0].detach().cpu().numpy()[0, :]
+        x = sound_generator.generate([scene])[0, :, :].detach().cpu().numpy()
         
         x /= np.abs(x).max() + 1e-7
         # is 16kHz - AUdiogen Fs