[demo] support English

app.py

@@ -18,18 +18,23 @@ import gradio as gr
 import wenetruntime as wenet
 import librosa
 
-
 wenet.set_log_level(2)
 decoder_cn = wenet.Decoder(lang='chs')
+decoder_en = wenet.Decoder(lang='en')
 
 
-def recognition(audio):
+def recognition(audio, lang='CN'):
     if audio is None:
         return "Input Error! Please enter one audio!"
     y, _ = librosa.load(audio, sr=16000)
     # NOTE: model supports 16k sample_rate
     y = (y * (1 << 15)).astype("int16")
-    ans = decoder_cn.decode(y.tobytes(), True)
+    if lang == 'CN':
+        ans = decoder_cn.decode(y.tobytes(), True)
+    elif lang == 'EN':
+        ans = decoder_en.decode(y.tobytes(), True)
+    else:
+        return "ERROR! Please select a language!"
     if ans is None:
         return "ERROR! No text output! Please try again!"
     # NOTE: ans (json)
@@ -42,20 +47,28 @@ def recognition(audio):
 
 
 # input
-inputs = gr.inputs.Audio(source="microphone", type="filepath", label='Input audio')
+inputs = [
+    gr.inputs.Audio(source="microphone", type="filepath", label='Input audio'),
+    gr.Radio(['EN', 'CN'], label='Language')
+]
 
 output = gr.outputs.Textbox(label="Output Text")
 
 examples = [
-   ['examples/BAC009S0767W0127.wav'],
-   ['examples/BAC009S0767W0424.wav'],
-   ['examples/BAC009S0767W0488.wav'],
+    ['examples/BAC009S0767W0127.wav', 'CN'],
+    ['examples/BAC009S0767W0424.wav', 'CN'],
+    ['examples/BAC009S0767W0488.wav', 'CN'],
+    ['examples/1995-1836-0002.flac', 'EN'],
+    ['examples/61-70968-0000.flac', 'EN'],
+    ['examples/672-122797-0000.flac', 'EN'],
 ]
 
 text = "Speech Recognition in WeNet | 基于 WeNet 的语音识别"
 
 # description
-description = ("Wenet Demo ! This is a Mandarin streaming speech recognition !")
+description = (
+    "Wenet Demo ! This is a speech recognition demo that supports Mandarin and English !"
+)
 
 article = (
     "<p style='text-align: center'>"