|
|
Post by Johnkenn on Jan 7, 2024 15:53:41 GMT -6
I rarely send out more than a 24 bit wav and mp3 to people. I had previously only used either when converting to 16 bits. I usually just use iTunes to make the mp3. Does that require dithering? Does the codec that does that handle it? Excuse my dumbassery
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 7, 2024 16:06:39 GMT -6
i always dither when going out to 24-bit. i make the mp3 in reaper from the dithered 16-bit master or for preview, just render the mix out as an mp3 because I always have a 24-bit dither (goodhertz good dither now) as my last insert.
|
|
|
|
Post by spindrift on Jan 7, 2024 16:45:52 GMT -6
In 32bit DAW sessions, I dither right before the converter outputs to my summing mixer or insert sends. Also, right before my stereo mix outputs to 2bus compression/EQ and recapture. If you’re bouncing from a 32bit session, dither to 24 bit for making 24 bit WAVs.
|
|
|
|
Post by Johnkenn on Jan 7, 2024 17:00:38 GMT -6
But if you're not changing bit rate, you don't have to dither, right...or at least I never do. Is there a need to put in dither if you plan on making an mp3 out of the 24 bit wav?
|
|
|
|
Post by ab101 on Jan 7, 2024 17:17:09 GMT -6
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 7, 2024 18:00:32 GMT -6
But if you're not changing bit rate, you don't have to dither, right...or at least I never do. Is there a need to put in dither if you plan on making an mp3 out of the 24 bit wav? dither is about preventing quantization error that results in distortion from changing bit depth, not the sample rate, of fixed point pcm samples or performing math with fixed point samples. the bottom of the samples is chopped off when the sample is rescaled, the next sample can be either lower or greater than it in value, there's no way to work it out, so there's distortion. proper audio dither replaces the bottom with 2 least significant bits worth of random noise because the distortion of random noise is random noise. this is triangular density probability function dither or tdpf dither. rectangular density is one significant bit of random noise and is insufficient. |
|
|
|
Post by viciousbliss on Jan 7, 2024 20:44:19 GMT -6
There's already at least one great dither discussion on this forum. The consensus was to dither before a hardware insert and when bouncing, right? Someone here posted about the Stillwell Bitter plugin to see how many bits you were sending out. In PT it said 32-bits. Once I put Maat Linpro in, it read 24-bit and I didn't need to insert a dither plugin in front of each hardware insert to maintain 24-bits. Sometimes I get weird feedback if I use the Maat too much, then I switch to Good dither. The Maat maintains the clarity a lot better than any other dither that I've tried. There's a guy developing a new plugin that posts on GS. I subscribe to the thread and I don't recall there being any updates in a while though.
|
|
Ari
Full Member
Posts: 35
|
Post by Ari on Jan 7, 2024 21:09:30 GMT -6
mp3 is not a PCM format to begin. you should encode straight from 32 or 64 bit float with no dither. the effective dynamic range of mp3 with pure tones is about 81 bit equivalent in pcm, though with more complicated signals it gets complicated. Still, it's far more than 16 bit and dithering the source will only hurt an MP3.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 7, 2024 21:41:08 GMT -6
mp3 is not a PCM format to begin. you should encode straight from 32 or 64 bit float. the effective dynamic range of mp3 with pure tones is about 81 bit equivalent in pcm, though with more complicated signals it gets complicated. Still, it's far more than 16 bit and dithering the source will only hurt an MP3. ?? |
|
Ari
Full Member
Posts: 35
|
Post by Ari on Jan 7, 2024 21:49:35 GMT -6
mp3 is not a PCM format to begin. you should encode straight from 32 or 64 bit float. the effective dynamic range of mp3 with pure tones is about 81 bit equivalent in pcm, though with more complicated signals it gets complicated. Still, it's far more than 16 bit and dithering the source will only hurt an MP3. ?? Hey Dan, what part of this did you find confusing? MP3 does not have a dynamic range in a traditional way. Each frame of the MP3 is stored somewhat analogous to floating point wav with an associated "global gain" that determines the gain of that frame. MP3 from a dithered source could affect the quality or final file size if it's VBR, or worst case scenario would be audible. Ideally, MP3 encoding and decoding both happen in float and the only dithering involved is applied after decoding the mp3, if the playback device needs it. |
|
|
|
Post by rowmat on Jan 7, 2024 22:03:57 GMT -6
Yeah time to stop dithering and just get on with it!
|
|
Ari
Full Member
Posts: 35
|
Post by Ari on Jan 7, 2024 22:11:03 GMT -6
THAT BEING SAID, a lot of programs pipe 24 bit into the mp3 encoder, and a lot of third party mp3 encoders only accept 24 bit, so there are some artificial limits on DR, but you shouldn't dither in this case either IIRC.
|
|
|
|
Post by Bob Olhsson on Jan 8, 2024 7:54:58 GMT -6
The whole idea of optional dither is as stupid as the idea of optional tape bias would be! Almost any digital signal processing kicks the bit depth up to infinity. The distortion will always be more audible than the dither. There simply is nothing to gain by not dithering. While dither can appear to have no benefit, not dithering can come back and bite you down the line.
|
|
|
|
Post by viciousbliss on Jan 8, 2024 8:07:11 GMT -6
The whole idea of optional dither is as stupid as the idea of optional tape bias would be! Almost any digital signal processing kicks the bit depth up to infinity. The distortion will always be more audible than the dither. There simply nothing to gain by not dithering. While dither can appear to have no benefit, not dithering can come back and bite you down the line. Bob, I've really appreciated all your research into dithering. It definitely does make a lot of difference. So much so that when people talk about the sound of hardware devices, I ask them if they've applied dither. The sound differences became more apparent to me the more devices were in a chain too. People who don't dither are having to base their whole style and workflow on reacting to the distortion caused by that practice as most everyone here must know already. Just something I wanted to point out for anyone reading who has never tried dither. I'm surprised UAD has never bothered with a dither plugin. The industry doesn't treat it with any importance. The best discussion on it has taken place on this board from what I remember. Digital and plugins are definitely hindered from reaching their full potential due to things like not dithering. It'd be great if Cedar came out with one. The dither in Adaptive Limiter 2 is quite good. Something where I notice the switch between 24 and 16 less than with others. Even converting something like the 24-bit file of Perhaps by Guns N Roses was iffy. Tried it with RX and the Maat in PT. I liked the 16-bit file a lot less. I'm always using the flat dither. Well, maybe that new guy's plugin will beat everything else on the market. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 8, 2024 8:20:34 GMT -6
THAT BEING SAID, a lot of programs pipe 24 bit into the mp3 encoder, and a lot of third party mp3 encoders only accept 24 bit, so there are some artificial limits on DR, but you shouldn't dither in this case either IIRC. You should dither to 24-bit because the 64/32 bit float to 24-bit fixed conversion causes distortion. So you can prevent that harmonic distortion (that is audible with rooms, reverbs, synths, and some higher gain guitar tones) at the expense of noise (that doesn't matter at all because it's below the johnson noise of all the equipment used). Just don't use shaped dither like Apogee UV22 which adds high frequency almost full scale tone with super aggressive shaping vs other algorithms like pow-r inside pro tools, logic, and weiss gear. the apogee can blow a tweeter. TDPF is best for 24-bit or anything that might be modulated or have digital volume control again unlike the cds in a cd player in a hifi system hitting a preamp. Not that this really matters because mp3 is so cooked anyway with tons of quantization, chirping, pre-echo etc. A rapper friend of mind can pick out mp3 vs .wav on earbuds from the pre-echo. |
|
|
|
Post by gravesnumber9 on Jan 8, 2024 11:32:30 GMT -6
In the history of dithering threads I have never read one that did not leave me more confused than when I started. And I've read a lot of them!
|
|
|
|
Post by noob on Jan 8, 2024 11:49:52 GMT -6
The whole idea of optional dither is as stupid as the idea of optional tape bias would be! Almost any digital signal processing kicks the bit depth up to infinity. The distortion will always be more audible than the dither. There simply nothing to gain by not dithering. While dither can appear to have no benefit, not dithering can come back and bite you down the line. Well now I have to start dithering religiously  |
|
|
|
Post by noob on Jan 8, 2024 11:58:57 GMT -6
I'm reading that Luna dithers automatically on export, which kind of makes sense. It doesn't have any option to do it like other DAWs I use.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 8, 2024 12:24:54 GMT -6
The whole idea of optional dither is as stupid as the idea of optional tape bias would be! Almost any digital signal processing kicks the bit depth up to infinity. The distortion will always be more audible than the dither. There simply nothing to gain by not dithering. While dither can appear to have no benefit, not dithering can come back and bite you down the line. Bob, I've really appreciated all your research into dithering. It definitely does make a lot of difference. So much so that when people talk about the sound of hardware devices, I ask them if they've applied dither. The sound differences became more apparent to me the more devices were in a chain too. People who don't dither are having to base their whole style and workflow on reacting to the distortion caused by that practice as most everyone here must know already. Just something I wanted to point out for anyone reading who has never tried dither. I'm surprised UAD has never bothered with a dither plugin. The industry doesn't treat it with any importance. The best discussion on it has taken place on this board from what I remember. Digital and plugins are definitely hindered from reaching their full potential due to things like not dithering. It'd be great if Cedar came out with one. The dither in Adaptive Limiter 2 is quite good. Something where I notice the switch between 24 and 16 less than with others. Even converting something like the 24-bit file of Perhaps by Guns N Roses was iffy. Tried it with RX and the Maat in PT. I liked the 16-bit file a lot less. I'm always using the flat dither. Well, maybe that new guy's plugin will beat everything else on the market. fixed point pcm scaling is linear with 2 lsb of tdpf dither. amount of noise in a dithered 16-bit audio is almost always beneath that of a quiet mic cranked up with a quiet pre. the issue is prosumer and consumer hifi converters that only spec out well noise/distortion floor wise (they still usually sound like crap unless it's something like a bricasti m1 or weiss hifi box) at close to 0 dbfs but not exceeding it or coming between -1 to -3 dbfs where they will start clipping. turn down these pieces of crap to +4 dbu / 0 vu or −10 dBV and they spec out like in spec analog gear but throw in junk or counterfeit parts or lack of qc, many of the units in the wild were never up to the spec
|
|
|
|
Post by ab101 on Jan 8, 2024 12:39:11 GMT -6
So here is where I am still somewhat confused.
1. Dither from 24 bit file to MP3?
2. If yes, then is it best to dither from 24bit to 16bit file and then convert 16bit file to mp3 (and with dither or not???)
3. And, is dither ever a bad thing? Assuming using a high grade dither. Wavelab 11 uses Lin pro dither.
|
|
|
|
Post by Bob Olhsson on Jan 8, 2024 12:51:39 GMT -6
I saw a thing on YouTube that showed a properly dithered 16-bit file would sound better if the player truncated to 16. Apparently, many do just that.
|
|
|
|
Post by gravesnumber9 on Jan 8, 2024 13:12:12 GMT -6
So here is where I am still somewhat confused. 1. Dither from 24 bit file to MP3? 2. If yes, then is it best to dither from 24bit to 16bit file and then convert 16bit file to mp3 (and with dither or not???) 3. And, is dither ever a bad thing? Assuming using a high grade dither. Wavelab 11 uses Lin pro dither. I'd like to add... 4. Dither before hitting processing hardware using tools like S1's Pipeline XT or Logic Pro's I/O plugin? (Looked everywhere in the S1 documentation and couldn't figure out if they apply their dither to Pipeline. Does it matter?) |
|
|
|
Post by bossanova on Jan 8, 2024 21:49:33 GMT -6
From Bob and Paul Frindle collectively, and probably others, I’ve been told to always Triangle dither when leaving the box for outboard processing, or, exporting to a lower bitrate, including 24-bit files that have been mixed in 32-bit float and will be going out to 16 *or* 24 for export.
|
|
kojei
New Member
Posts: 7
|
Post by kojei on Jan 9, 2024 8:31:41 GMT -6
Some MP3 encoders can convert without truncating the floating point. When using such encoders, even files exceeding 0dBFS can be converted without dithering, allowing the creation of MP3 files without clipping. Of course, using dithering will truncate parts exceeding 0dBFS, resulting in clipping. Below, I have prepared two MP3 files for demonstration. One is a file converted without dithering, and the other is converted with a 16-bit fixed dither. When the non-dithered file is loaded into a DAW and observed with an oscilloscope, the sine wave is preserved, whereas the file with dithering shows clipping. However, I believe not all DAWs are compatible with this. (AFAK Reaper, RX, Audition are OK) files here : u.pcloud.link/publink/show?code=kZ23zX0Z8LbbWJAIg3SSP943HitkJ4HHWqt7 |
|
|
|
Post by Johnkenn on Jan 9, 2024 10:40:29 GMT -6
In the history of dithering threads I have never read one that did not leave me more confused than when I started. And I've read a lot of them!  |
|
