Programming Audio is Dangerous for Your Hearing!

Streamed Live on Twitch: / tsoding
Panim Playlist: • Panim
- 00:00:00 - Safe zone
- 02:29:28 - LOUD SOUND WARNING
- 02:29:38 - Safe zone
References:
- Tsoding - Musializer github.com/tsoding/musializer
Support:
- BTC: bc1qj820dmeazpeq5pjn89mlh9lhws7ghs9v34x9v9
- Pay for my VPS: zap-hosting.com/en/shop/donat...

[@TsodingDaily]

02:29:28 - LOUD SOUND WARNING

[@jayshartzer844]

What? I can't hear you over the loud sound

[@Eugensson]

Clicked the timestamp without hesitation... dunno what i was expecting 💀

[@davidensikat4707]

kzbin.info/www/bejne/iJq7ZXl3hqhsmck

[@seventhtenth]

a programmer

[@generationalgamers8925]

Samples and frames are the same I thought

[@duality4y]

when developing audio never put the headset on your ears ... it's something you learn quickly ...

[@TheMarcsOv]

i once stunned myself so hard with a full amplitude square wave that i got physically dizzy for like 2 minutes

[@U20E0]

oof. Why can headphone speakers even be that loud

[@benjaminquiedeville4476]

As an audio dev I feel that pain I always clip the output signal between -1.0 and 1.0 BEFORE the master volume parameter. It saved me a lot of time

[@benjaminquiedeville4476]

And in debug mode, put some assert(abs(x)

[@caspianmerlin6434]

Hey, any tips for resources for learning audio dev concepts?

[@berndeckenfels]

You can also create many more file descriptors (pipe on fd3) And either use ffmpeg syntax pipe:3 or in Linux /proc/self/fd/3 Advantage is you don’t need with leftover fifos or create unique names for parallel run

[@mikoposter]

:3

[@z4pp820]

I usually route the output signal into a limiter with a brickwall output at -6 db. Has saved my ears more than once

[@ziv132]

When you are rendering the video you have args in the output for an audio codec (that are unused since it's only video) Also you might want to add those options to when you are encoding the audio. You can also leave it as a wav and do the encoding when you merge the video and audio together, then you would put the encoding for the audio instead of the copy in -c:a copy

[@berndeckenfels]

That’s literally what the whole video was about

[@Tanger68]

Skipped straight to the loud sound with confidence that it couldn't be that loud. It's quite loud.

[@rogo7330]

Since you execve ffmpeg you could just use pipe() to create two new file descriptors that reffer to input and output of the same pipe. After forking and executing ffmpeg you just say it to read from pipe: and on your end you write to input fd. File descriptors, if nothing special was done to them, just pass-on to the next process, execve does not closes them either.

[@rogo7330]

Already discussed in the video. (crawling back into the dark corner in shame)

[@blackhaze3856]

How a struct sounds?

[@drdca8263]

How long have the captions included what was said by twitch chat?

[@ikeofili355]

I don't know if you want to hear this but you would need 2 different pipes (and maybe use 2 file descriptors for each ffmpeg instance to read from). The error happens because you only have one p->pipe and it is overwritten by `start_ffmpeg_audio`. So then you send video to the audio pipe. You would probably need a p->video_pipe and p->audio_pipe

[@berndeckenfels]

He has two (fifo) but one blocks if the other side not opens them in right order (or at all). But I think he could also use two pipes which don’t have the blocking problem, with fem pegs syntax, pipe:3 should work.

[@berndeckenfels]

47:58 it’s worse since it overcommits you can allocate more than ram+swap. But there might be soft limits

[@atiedebee1020]

I can vouch for this, I once tried doing audio encoding, and when you create a bug with that shit, youll remember it

[@sanjaux]

It really is!

[@satchelfrost6531]

Not sure if you mentioned this, but why is panim not open source like many of your other projects? The only reason I ask is because I like to go back and forth between watching the video and looking at the final code. It really helped me understand the ml-notes stream.

[@user-uf4rx5ih3v]

He releases all of his projects on his github page when he feels they're in a finished state for the time being. He rarely works on something for more then a month so expect the code soon.

[@user-cx6ec2kp6u]

Knowing when you are rendering can be used to show a low res preview and render in high res 17:38

[@anilgr1111]

Yes we are alive !

[@adrianmanea]

What do you use to zoom in and highlight your mouse pointer while streaming? 1:18:49

[@victordvickie]

tsoding/boomer github

[@mkjyt1]

Hey Tsoding!

[@c4llv07e]

The daily use of OpenBSD was interesting, but sometimes I had to install Linux for some software. I miss these man pages.

[@brambasieldev]

Unsafe C programmer got cocky and flew too close to the sun. Rust borrow checker would have prevented this.

[@henrikholst7490]

Totally! Btw. I run Arch.

[@boccobadz]

Rust would prevent shipping any usable product.

[@brambasieldev]

@@boccobadz me with c++ 😂 (i use c++ btw)

[@heejadeetorn753]

Rust, the word "unsafe" and Arch in the same comment thread. The only thing missing is a person of 40% sui rate

[@apenaswellignton]

Pog

[@ratatouillegamer7132]

i'm fear about hear, what is the problem? if i hear with headphones is a problem?

[@multiarray2320]

what are you talking about?

[@blastygamez]

Oh no...

[@anidnmeno]

the fact that there's an ffmpeg Twitter

[@RandomGeometryDashStuff]

if you only encode video, why not just use x264 instead of ffmpeg?

[@seventhtenth]

25:20 WHAT did he JUST SAY

[@marshall00721]

yo dad wassupp

[@berndeckenfels]

„When we have one pipe we can have two, nobody is stopping us“- unless the child opens them in wrong sequence :) (I would make open nonblocking)

[@binitrupakheti4246]

Thats why u should've used rust

[@Eugensson]

Why?

[@juan_cb3357]

It pretends to be safe ​@@Eugensson