Extra latency in comparison to Web Audio Api

[YoniH]

I had a feeling that tone.js adds extra latency on top of the standard web audio api, so I wrote a small app to test it:

https://yonih.github.io/latency-test/

You can find the code here.

At least in my setup (Mac M1, Chrome browser) the difference between plain Web Audio Api and Tone.js is very audiable and unacceptable.

I recorded myself playing with the test app, so I can look at the recording's waveform and measure exactly how much time it took between the keystroke sound and the time that the sample sound came out of the speakers. That is, the latency. Here is the recording.

And here are the results: Web audio api: 76ms 61ms 57ms Tone.js: 157ms. 162ms, 160ms

Am I missing something? Did I write the tone.js code wrong? Can I expect tone.js to respond as quickly as plain web audio api?

[YoniH]

Looking at the results, it seems that the difference is about 100ms, which is exactly the default lookahead. I did set the the lookahead to 0 though, and also called Tone.immediate(). Maybe I have a bug around that area?

[marcelblum]

You're right that the cause of the latency is lookAhead. But in your code you're creating your Players before instantiating a new audioContext with a lower lookAhead. That results in those players still using the older default Tone context with the higher lookAhead. You actually don't need to create a new context if all you want is to lower the lookAhead, that can be changed dynamically at any time Tone.getContext().lookAhead = 0.

[YoniH]

Yes, that was it 🎉🎉🎉 Now the latency I'm getting is similar to plain Web Audio Api. Thanks 😄