Update app.py

app.py

@@ -16,7 +16,7 @@
 # limitations under the License.
 # -------------------------------------------------------------------------
 
-#import spaces
+import spaces
 import gradio as gr
 from PIL import Image
 #from pydub import AudioSegment
@@ -64,11 +64,11 @@ SIDEBAR_INFO = f"""
 """
 
 # ------------transcribe section------------
-device = "cuda" if torch.cuda.is_available() else "cpu"
-pipe = pipeline("automatic-speech-recognition", model="NbAiLab/nb-whisper-large", device=device) #  chunk_length_s=30, generate_kwargs={'task': 'transcribe', 'language': 'no'}
+) #  chunk_length_s=30, generate_kwargs={'task': 'transcribe', 'language': 'no'}
 
-#@spaces.GPU()
-def transcribe(microphone, file_upload, batch_size=15):
+@spaces.GPU()
+def transcribe(microphone, file_upload):
+    #--------------____________________________________________--------------"
     warn_output = ""
     if (microphone is not None) and (file_upload is not None):
         warn_output = (
@@ -80,9 +80,16 @@ def transcribe(microphone, file_upload, batch_size=15):
         return "ERROR: You have to either use the microphone or upload an audio file"
 
     file = microphone if microphone is not None else file_upload
+    #--------------____________________________________________--------------"
 
     start_time = time.time()
-    text = pipe(file, batch_size=batch_size, return_timestamps=False)["text"]
+    
+    #--------------____________________________________________--------------"
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    pipe = pipeline("automatic-speech-recognition", model="NbAiLab/nb-whisper-large", device=device)
+    
+    text = pipe(file)["text"]
+    #--------------____________________________________________--------------"
 
     end_time = time.time()
     output_time = end_time - start_time
@@ -93,10 +100,8 @@ def transcribe(microphone, file_upload, batch_size=15):
     gpu_utilization, gpu_memory = GPUInfo.gpu_usage()
     gpu_utilization = gpu_utilization[0] if len(gpu_utilization) > 0 else 0
     gpu_memory = gpu_memory[0] if len(gpu_memory) > 0 else 0
-
     # --CPU metric
     cpu_usage = psutil.cpu_percent(interval=1)
-
     # --system info string
     system_info = f"""
     *Memory: {memory.total / (1024 * 1024 * 1024):.2f}GB, used: {memory.percent}%, available: {memory.available / (1024 * 1024 * 1024):.2f}GB.*
@@ -106,8 +111,7 @@ def transcribe(microphone, file_upload, batch_size=15):
     *CPU Usage: {cpu_usage}%*
     """
 
-    return warn_output + text.strip(), system_info
-
+    return warn_output + text, system_info
 
 def save_to_pdf(text, summary):
     pdf = FPDF()
@@ -138,7 +142,7 @@ with iface:
 
     with gr.Row():
         gr.Markdown('''
-        ##### 1. Last opp lydfila 🔊 (dropp ’drag-n-drop’)
+        ##### 1. Last opp lydfila 🔊 
         2. Trykk på "Transkriber" knappen og vent på svar ☕️
         3. Går rimelig bra kjapt med Norwegian NB-Whisper Large⚡️
         4. Planlegger tilleggs-funksjoner senere😅