Update app.py

app.py

@@ -54,6 +54,7 @@ def recognize(audio: tuple[int, np.ndarray], models, language):
         results = []
         for name, model in models.items():
             if length > 20 and name == "alphacep/vosk-model-small-ru":
+                gr.Warning(f"Model {name} only supports audio no longer than 20 s.")
                 continue
             start = timer()
             result = model.recognize(waveform, sample_rate=sample_rate, language=language)
@@ -110,7 +111,7 @@ with gr.Blocks() as recognize_short:
 
 
 with gr.Blocks() as recognize_long:
-    gr.Markdown("For better results, you need to adjust the VAD parameters.")
+    gr.Markdown("The default VAD parameters are used. For best results, you should adjust the VAD parameters in your app.")
     name = gr.Dropdown(models_vad.keys(), label="Model")
     audio = gr.Audio(min_length=1, max_length=300)
     with gr.Row():
@@ -135,6 +136,13 @@ with gr.Blocks(title="onnx-asr demo") as demo:
     # ASR demo using onnx-asr
     **[onnx-asr](https://github.com/istupakov/onnx-asr)** is a Python package for Automatic Speech Recognition using ONNX models.
     The package is written in pure Python with minimal dependencies (no `pytorch` or `transformers`).
+
+    **onnx-asr** is very easy to use (see [Readme](https://github.com/istupakov/onnx-asr?tab=readme-ov-file) for more examples):
+    ```py
+    import onnx_asr
+    model = onnx_asr.load_model("nemo-parakeet-tdt-0.6b-v2")
+    print(model.recognize("test.wav"))
+    ```
     """)
     gr.TabbedInterface(
         [recognize_short, recognize_long],