Update app.py

app.py

@@ -19,7 +19,7 @@ custom_repo_id = "AkhilTolani/vocals-english"
 
 custom_model = ParlerTTSForConditionalGeneration.from_pretrained(custom_repo_id).to(device)
 
-tokenizer = AutoTokenizer.from_pretrained(custom_repo_id)
+tokenizer = AutoTokenizer.from_pretrained("parler-tts/parler_tts_mini_v0.1")
 
 SEED = 456
 
@@ -168,12 +168,6 @@ class ParlerTTSStreamer(BaseStreamer):
 sampling_rate = custom_model.audio_encoder.config.sampling_rate
 frame_rate = custom_model.audio_encoder.config.frame_rate
 
-def calculate_min_length(text, words_per_beat=2.0, beats_per_second=3.16):
-    words = len(text.split())
-    total_beats = words / words_per_beat
-    min_length_seconds = total_beats / beats_per_second
-    return int(min_length_seconds)
-
 @spaces.GPU
 def generate_base(text, description, play_steps_in_s=2.0):
     play_steps = int(frame_rate * play_steps_in_s)
@@ -182,13 +176,11 @@ def generate_base(text, description, play_steps_in_s=2.0):
     inputs = tokenizer(description, return_tensors="pt").to(device)
     prompt = tokenizer(text, return_tensors="pt").to(device)
 
-    min_length_seconds = calculate_min_length(text)
-
     generation_kwargs = dict(
         input_ids=inputs.input_ids,
         prompt_input_ids=prompt.input_ids,
         streamer=streamer,
-        min_length=min_length_seconds,
+        min_length=20,
     )
 
     set_seed(SEED)