add examples

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*wav filter=lfs diff=lfs merge=lfs -text

app.py

@@ -82,7 +82,7 @@ def transcribe(inputs):
         return result
 
     distil_pipe._forward = _forward_distil_time
-    distil_text = distil_pipe(inputs, batch_size=BATCH_SIZE)["text"]
+    distil_text = distil_pipe(inputs.copy(), batch_size=BATCH_SIZE)["text"]
     yield distil_text, distil_runtime, None, None, None
 
     def _forward_time(*args, **kwargs):
@@ -99,6 +99,9 @@ def transcribe(inputs):
     yield distil_text, distil_runtime, text, runtime
 
 if __name__ == "__main__":
+    # CUDA warm-up step so that we don't count it in our examples
+    _ = distil_pipe("./assets/example_1.wav")
+
     with gr.Blocks() as demo:
         gr.HTML(
             """
@@ -136,4 +139,12 @@ if __name__ == "__main__":
             inputs=audio,
             outputs=[distil_transcription, distil_runtime, transcription, runtime],
         )
+        gr.Markdown("## Examples")
+        gr.Examples(
+            [["./assets/example_1.wav"], ["./assets/example_2.wav"]],
+            audio,
+            outputs=[distil_transcription, distil_runtime, transcription, runtime],
+            fn=transcribe,
+            cache_examples=True,
+        )
     demo.queue(max_size=10).launch()

assets/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e938b9f81dea096ec7d3752e90afca8d370f7a461d3a08e1a559f4440ed055d
+size 1963810

assets/example_2.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81fc0857f7fe11416ede431db713a02fdb787bbc049802fe74c791f3b44e5bf4
+size 1920044