How to lengthen the Whisper max audio length?

[stinoga]

Question

I'm working from the webgpu-whisper demo, and I'm having a hard time lengthening the maximum audio input allowed. I made the following changes:

-const MAX_AUDIO_LENGTH = 30; // seconds
+const MAX_AUDIO_LENGTH = 120; // seconds

-const MAX_NEW_TOKENS = 64;
+const MAX_NEW_TOKENS = 624;

This seems to allow for longer input, but after 30 seconds I get the following error:

Attempting to extract features for audio longer than 30 seconds. If using a pipeline to extract transcript from a long audio clip, remember to specify `chunk_length_s` and/or `stride_length_s`.

I can't seem to find where to add stride_length_s in the demo code, however. Could someone point me in the right direction?