asr

app.py

@@ -9,7 +9,7 @@ mms_transcribe = gr.Interface(
         gr.Audio(),
         gr.Dropdown(
             choices=[model for model in asr.models_info] + ["Compare All Models"],
-            label="Select Model for ASR",
+            label="Select a Model for ASR",
             value="ixxan/wav2vec2-large-mms-1b-uyghur-latin",
             interactive=True
         ),
@@ -18,12 +18,11 @@ mms_transcribe = gr.Interface(
         gr.Textbox(label="Uyghur Arabic Transcription"),
         gr.Textbox(label="Uyghur Latin Transcription"),
     ],
-    #examples=ASR_EXAMPLES,
+    examples=util.asr_examples,
     title="Speech-to-text",
     description=(
         "Transcribe Uyghur speech audio from a microphone or input file."
     ),
-    #article=ASR_NOTE,
     allow_flagging="never",
 )
 
@@ -33,7 +32,7 @@ mms_synthesize = gr.Interface(
         gr.Text(label="Input text"),
         gr.Dropdown(
             choices=[model for model in tts.models_info],
-            label="Select Model for TTS",
+            label="Select a Model for TTS",
             value="Meta-MMS",
             interactive=True
         )
@@ -43,7 +42,9 @@ mms_synthesize = gr.Interface(
     ],
     examples=util.tts_examples,
     title="Text-to-speech",
-    description=("Generate audio from input text."),
+    description=(
+        "Generate audio from input Uyghur text."
+        ),
     allow_flagging="never",
 )
 

util.py

@@ -4,6 +4,9 @@ from umsc import UgMultiScriptConverter
 ug_arab_to_latn = UgMultiScriptConverter('UAS', 'ULS')
 ug_latn_to_arab = UgMultiScriptConverter('ULS', 'UAS')
 
+asr_examples = [['examples/1.wav', 'ixxan/wav2vec2-large-mms-1b-uyghur-latin'],
+                ['examples/2.wav', 'ixxan/wav2vec2-large-mms-1b-uyghur-latin']]
+
 tts_examples = [
     ["مەكتەپكە بارغاندا تېخىمۇ بىلىملىك بولۇمەن.", "Meta-MMS"],
     ["قىلىچ قان تامغۇزسا، بەگ ئەل ئالىدۇ؛ قەلەمدىن سىياھتانسا، ئالتۇن كېلىدۇ.", "Meta-MMS"],