Update app.py

app.py

@@ -52,7 +52,7 @@ vocos = Vocos.from_pretrained("charactr/vocos-mel-24khz")
 target_sample_rate = 24000
 n_mel_channels = 100
 hop_length = 256
-target_rms = 0.15
+target_rms = 0.1
 nfe_step = 8  # 16, 32
 cfg_strength = 2.0
 ode_method = "euler"
@@ -274,13 +274,7 @@ def infer(ref_audio_orig, ref_text, gen_text, exp_name, remove_silence, cross_fa
 
     if not ref_text.strip():
         gr.Info("No reference text provided, transcribing reference audio...")
-        ref_text = pipe(
-            ref_audio,
-            chunk_length_s=30,
-            batch_size=128,
-            generate_kwargs={"task": "transcribe"},
-            return_timestamps=False,
-        )["text"].strip()
+        ref_text = gen_text
         gr.Info("Finished transcription")
     else:
         gr.Info("Using custom reference text...")