cinera_handmade.network/cmuratori/hero/code/code020.hmml

63 lines
4.1 KiB
Plaintext
Raw Normal View History

[video member=cmuratori stream_platform=twitch stream_username=handmade_hero project=code title="Debugging the Audio Sync" vod_platform=youtube id=hELF8KRqSIs annotator=jacebennett annotator=Miblo annotator=garlandobloom]
[2:44][Fixing an off by one error]
[5:12][Explanation of off by one errors]
[7:13][Overview of the state of our audio code]
[11:05][A breakdown of how time is spent during each frame]
[14:07][A possible, though not great, solution to audio latency]
[16:22][Tradeoff between audio lag and input lag]
[17:51][Falling back to a little bit stupider]
[21:58][Beginning to change the code for the audio output method]
[22:48][Finding the minimum expected audio latency]
[27:31][Converting to seconds using dimensional analysis]
[31:34][Writing sound based on the audio latency]
[34:14][Great aside about keeping code fluid and a bit messy in the early stages]
[37:00][Determining the position to write the audio to]
[46:27][How to deal with the low latency case]
[51:16][A puzzling situation]
[53:00][The aha moment]
[55:34][The approach we will be taking for low latency]
[57:36][The higher latency approach]
[1:01:14][Another snag]
[1:07:08][Putting our approach down in words]
[1:13:02][Implementing the two case latency handling]
[1:18:34][Do over due to deleted code]
[1:20:08][Putting in the latency branching code]
[1:23:35][Determining if the sound card is latent]
[1:26:02][Mapping the WriteCursor and the PlayCursor into the same space]
[1:27:51][Overview of the code determining how to write the sound]
[1:30:00][Cleaning up]
[1:31:22][How can we possibly predict how many bytes the game should write?]
[1:33:02][Getting sound output separately with GameGetSoundSamples()]
[1:36:45][More cleaning up]
[1:37:43][Computing SafetyBytes]
[1:39:06][Implementing GameGetSoundSamples() in the game code]
[1:40:10][Testing is surprisingly successful, no bugs? NO WAY!]
[1:42:47][Improving our sound output debug drawing]
[1:48:25][Recording some additional details in our markers]
[1:56:36][An assertion fires and some cleanup]
[1:58:33][First look at the debug drawing output and adding global pause key]
[2:02:38][Taking a good look at the debug drawing output]
[2:04:40][A couple early questions]
[2:05:47][Final Thoughts?]
[2:06:42][Drawing another debug marker]
[2:10:53][Improving our estimate of the PlayCursor position to reduce latency]
[2:17:59][Taking a look at how our estimate lines up]
[2:19:13][Drawing our 480 sample jitter window]
[2:22:15][Q&A]
[2:22:58][Did you try the audio without OBS?]
[2:24:15][Is it inexpensive to query for the play cursor and the write cursor?]
[2:25:03][Audio seems to be the most complex part until now, which will be the next most complicated thing?]
[2:26:58][Audio asserts are a pain when you're debugging other things. Change to log messages.]
[2:27:46][If we had a buffer that played from start to finish, would it be a valid syncing method to manually set AudioTimer = VideoTimer?]
[2:28:45][Fully visualize waveforms and remove gradient. Turn entire thing into massive audio visualizer.]
[2:29:42][Could you please increase ToneHz to 512, my crappy laptop speakers can't output 256?]
[2:30:29][Will this audio system be able to handle music and sound, or will the game have to mux the audio for us?]
[2:31:08][How will you make sure that you don't have regressions in some parts that are not that easy to spot right away?]
[2:31:54][I'm sorry if I'm being slow, but I still don't truly understand why exactly the audio is jumping. It just seems like we should be able to handle it more simply.]
[2:32:39][Why is the pitch of the sound changing after the game has been running a little while?]
[2:39:55][Callback to earlier half of question: (I'm sorry if I'm being slow, but I don't truly understand why the audio is skipping.) The original audio skipped when we moved the latency to be inside of three frames.]
[2:42:25][If it's not too much of an issue, can we have line numbers on the side when you are coding so we can get a grip of where you are.]
[2:43:08][It's more likely to be the accumulation of a small value onto a larger value that is a numerical problem. Mantissas not lining up.]
[2:44:03][Wrap-up]
[/video]