Update app.py

app.py

@@ -1,11 +1,10 @@
-import librosa
-import gradio as gr
-#from transformers import Wav2Vec2Tokenizer, Wav2Vec2ForCTC
 from transformers import pipeline
+import gradio as gr
+import librosa
 
 #Loading the model and the tokenizer
 model_name = "pgilles/wav2vec-xls-r-Luxembourgish20-with-LM"
-pipe = pipeline("automatic-speech-recognition", model=model_name)
+p = pipeline("automatic-speech-recognition", model=model_name)
 
 #tokenizer = Wav2Vec2Tokenizer.from_pretrained(model_name)
 #model = Wav2Vec2ForCTC.from_pretrained(model_name)
@@ -26,13 +25,24 @@ def load_data(input_file):
   return speech
     
 def asr_pipe(input_file):
-  transcription = pipe(input_file, chunk_length_s=3, stride_length_s=(1, 1))
+  transcription = p(input_file, chunk_length_s=3, stride_length_s=(1, 1))
   return transcription
   
-
-gr.Interface(asr_pipe,
-             inputs = gr.inputs.Audio(source="microphone", type="filepath", optional=True, label="Hei kënnt Dir Är Sprooch iwwert de Mikro ophuelen"),
-             outputs = gr.outputs.Textbox(label="Output Text"),
+def transcribe(audio, state=""):
+    #time.sleep(2)
+    text = p(audio)["text"]
+    state += text + " "
+    return state, state
+    
+gr.Interface(transcribe,
+             inputs = [
+               gr.inputs.Audio(source="microphone", type="filepath", optional=True, label="Hei kënnt Dir Är Sprooch iwwert de Mikro ophuelen"), 
+               "state"
+             ],
+             outputs = [
+               gr.outputs.Textbox(label="Erkannten Text"),
+               "state"
+             ],
              title="Sproocherkennung fir d'Lëtzebuergescht @uni.lu",
              description = "Dës App convertéiert Är geschwate Sprooch an de (méi oder manner richtegen ;-)) Text!",
              examples = [["ChamberMeisch.wav"], ["Chamber_Fayot_2005.wav"], ["Erlieft-a-Verzielt.wav"], ["Schnessen_Beispill.wav"]], theme="default").launch()