Update app.py

app.py

@@ -2,8 +2,10 @@ import gradio as gr
 from transformers import pipeline
 import numpy as np
 import librosa
+from punctuators.models import PunctCapSegModelONNX
 
 transcriber = pipeline("automatic-speech-recognition", model="Oysiyl/w2v-bert-2.0-ukrainian-colab-CV16.0")
+punct_cap_model = PunctCapSegModelONNX.from_pretrained("1-800-BAD-CODE/xlm-roberta_punctuation_fullstop_truecase")
 
 def transcribe(audio):
     sr, y = audio
@@ -11,8 +13,9 @@ def transcribe(audio):
     y /= np.max(np.abs(y))
     if sr != 16000:
         y = librosa.resample(y, orig_sr=sr, target_sr=16000)
-
-    return transcriber({"sampling_rate": 16000, "raw": y})["text"]
+    transcribed_text = transcriber({"sampling_rate": 16000, "raw": y})["text"]
+    punct_cap_text = punct_cap_model.infer(texts=[ukr_text], apply_sbd=True)[0][0]
+    return punct_cap_text
 
 
 demo = gr.Interface(