How to make square,triangle,saw waves is just add sine wave harmonics to the sine wave of the frequency you want your wavetable wave to be until you think the next harmonic is going to get past 20k hz
Even 8 years on this video is fantastic. I think an auditory example of the effect of dither would have been nice, but the excellent visual examples more than sufficed.
The catch here is that RU-vid encoding only leaves 16-bit audio, and 17-24-bit sweeps are rendered at the same level as 16-bit. Here's a workaround: Go to this blog post, which has the uncompressed audio in a player beneath this video, so you can play both in a single window: earlevel.com/main/2022/04/10/how-many-bits-can-you-hear-video/; as noted on the page, you just need to start the audio player two seconds into the video (when the green meter appears, as the sound starts), and mute the video's audio.
Well yeah if you're toggling the least significant bit with a fixed full scale we're just losing 6db for each bit added to the resolution. You're just proving that 16 bit is enough to cover the human sensitivity range (from effectively complete silence to painful to hear)
I don't get it. What exactly is the signal here? A sine sweep? And why is it getting quieter with rising bit depth? Do you add the bits "on top" so that the volume effectively *_halves_* with every additional bit? I don't get it.
Click "show more", in the info above to get a little more detail, and a full explanation of the signal is at www.earlevel.com/main/2013/03/24/perspective-on-dither. The signal is a digital square wave, you can see the harmonics sweep down in the scrolling spectrogram. Two main reasons why I went with a digital square wave over sine sweeps: First, I want it to be easier to hear-I'm not trying to MAKE it difficult to hear, I want the listener to have a sense that this easy-to-hear signal becomes difficult to hear sooner than they might have thought. Second, a sine would have progressively worse signal-to-noise ratio, and require dithering, whereas the digital square sweep is alway exact at any bit depth. Thanks for being the first comment!
It shows you effectively how low the noise floor is, based off 0 dbfs. For example, 16-bit at -90.3 dBFS means you could set your speakers to play at 90.3 dB above the noise floor (in a very quiet room say 30 dB). This would mean the loudest parts would be 120.3 dB (painfully loud, most home audio or headphones won’t get close to this) with the noise floor effectively inaudible. The demo shows that anything above 16-bit for playback is kinda pointless however higher bit depths are very useful in mixing and mastering the final output. Gives you extra headroom before bringing the final output down to a more reasonable 16 bits per sample.
Bro thank you so much. This is an AMAZING explanation of dither, you are an amazing teacher. Turning such a complicated subject into something easy to understand is insanely difficult, yet you managed to do it. Thank you
Thank you so much, it helps to know-be sure to subscribe if you haven't so you don't miss my (long delayed, but I won't let another year go by) intuitive explanation of sampling theory...
Thanks! You should be able to reach the website at earlevel.com...it's probably because the link has "http", from back before the site was secure-only. I'll update it...
so in in 24bit recording, rendering a 24 bit mix, I shouldn't need dither? and here is a question, if i used a 16bit sample in a 24bit recording, is their any steps reverse order here, to avoid issues
@@TEN-TIMES-HARDER It never hurts to dither 24-bit truncations, but it also won't even be heard. Now, some would say, "why not just said always dither, and make it easy on everyone?". The reason I make this point is because some feel that every external send in your DAW (out to a hardware reverb, compressor, tube EQ...) must be properly dithered or you project will be ruin. This is an unnecessary complication-besides the extra setup, there is the paranoia of making sure there are no further gain changes after the truncation, etc. And it's for something that impossible to hear. The truncation distortion of at 24 bits is far below the noise floor of any electronics possible (you can't escape Johnson noise and shot noise). If that weren't enough-for people who aren't electrical engineers, and doubt that fundamental truth-your ears can't hear it (people get fooled by reading that the ear has 140 dB of dynamic range-they forget that you will have permanent hearing damage if you exercise that extreme for even a few seconds on the loud end, and on the other end you're limited by the inherent noise in your room, noise of blood pumping through your veins, and the minimum energy required to deflect your eardrum (it's not massless). If that were enough, only the most artificial cases of computed music won't already have far more than enough of a noise floor to self-dither, as you say. The bottom line is that it's unhearable for several fundamental reasons.
Nice explanation. Although it's not centered around graphics I would add that pure randomness is the "naive" way when dithering an image. There are other less random algorithms that make the perceived image quality much better by placing the pixels in specific patterns. You briefly touched on this point for audio at the end.
Short version: Limited resolution (sample size) results in a grid of possible values. The grid can possibly result in patterns in certain audio material that sound like the audio is distorted. By jiggling the grid a tiny bit, randomly, we don't get error patterns that line up with the audio signal-instead of a distorted signal, it sounds like the clean signal plus a tiny bit of hiss. Analogy: Close one eye and spread your fingers in front of your view, and read this. Your fingers will block some words. Jiggle your hand side to side, quickly, a small amount. Now you can read all the words, even though there is a little blur in front of them. The blur is better than the blocking.
Thank you so much! I was freaking out as I render 24 bit wav and FL Studio kinda suggests there's no need to dither at 24 bit but this clarification helped me loads as I've already rendered atleast 50 tracks without dither (24 bit)
The visual analogy at 0:35 with pixels has blown my mind 🤯. I've been wondering what dithering is for about 10 years now, it always seemed something too difficult to understand for a regular musician, but thanks to you, I know what I'm doing now haha.
Brilliant. Much respect, Nigel. Your way of explaining these three 'techniques' in light of the underlying principle makes for an excellent understanding of signals and systems. Your talent may be needed in schools and stuff. Thank you!!
Hi Nigel, thank you for the video. Like the articles on earlevel.com I like how you use a strong theoretical approach as foundation and work your way up to the practical applications. Much appreciated and looking forward to more videos. Thanks for investing the time and energy to share knowledge. I recently became (re-)interested in DSP.
I just released my own dithering and bit reduction plug-in. It can be found at github.com/datajake1999/C1Bitcrusher BTW, Can you please upload the song you used in the video? Thanks.
It was an unfinished snippet of a song in progress, pretty arbitrary. I actually wanted to use something professional mastered, a couple of professional friends had said they could give me something. But I think they got nervous that my video would somehow highlight faults, or something-didn't happen.
If it's your own high-quality lossless project, will it make most sense not to dither if converting to lower sampling rate/bitrate (example: to AAC/MP3)? Because it came from a high-quality source, so it doesn't need to interpret anything else.
Good question. The process of dithering definitely should be the responsibility of the conversion from lossless, though. Otherwise, the error in the lossy format is not a linear function, so it's hard to predict the effect relative to truncation error. But it wouldn't make sense to dither first, or after. I did a quick look on the web, and Waves says explicitly, "Don’t dither before converting to MP3 or AAC". You're right going directly from a high-quality source.
