Greetings.
I'm curious about the possibility of building an all-digital DJ mixer from scratch. The goal would be to design, test, and fabricate everything from the ground up (within reason) to get the mixer that no one is ever actually going to build for me.
I'm sold on going all-digital for reasons I'm sure you guys are aware.
And I have access to a maker space with some electronics gurus who know a lot more about it than I do. I'm vaguely aware of the complexity of programming the DSP and having boards etched, and I have ideas about input and output stages and ADC/DAC.
My question is this: how hard would it be to integrate a computer audio interface (ideally using Apple CoreAudio)? The goal would be to plug in one USB or FW connection and have access to ~10 mono signals from the computer and sending either 2 or 4 mono signals back from the mixer to the computer with very low latency (<10ms).
The reason I'd want to do it that way is that the DJ software I use and plan to use (Traktor Pro, Traktor Pro 2 and/or Ableton Live) all use representations (either 32-bit float or 32-bit PCM) significantly "larger" than the 24-bit integers that most audio interfaces use. It seems worthwhile to simply maintain that representation if possible all the way through the mixer to the final DAC before an analog gain stage at the final output and not waste time or incorporate flaws from DA/AD conversion if it's not necessary.
Is that even possible? Projects like soundflower and jack exist, so that leads me to believe that it is possible, but I have no idea how to do the hardware side even if I did manage to wrap my head around the code.
Thanks.
I'm curious about the possibility of building an all-digital DJ mixer from scratch. The goal would be to design, test, and fabricate everything from the ground up (within reason) to get the mixer that no one is ever actually going to build for me.
I'm sold on going all-digital for reasons I'm sure you guys are aware.
And I have access to a maker space with some electronics gurus who know a lot more about it than I do. I'm vaguely aware of the complexity of programming the DSP and having boards etched, and I have ideas about input and output stages and ADC/DAC.
My question is this: how hard would it be to integrate a computer audio interface (ideally using Apple CoreAudio)? The goal would be to plug in one USB or FW connection and have access to ~10 mono signals from the computer and sending either 2 or 4 mono signals back from the mixer to the computer with very low latency (<10ms).
The reason I'd want to do it that way is that the DJ software I use and plan to use (Traktor Pro, Traktor Pro 2 and/or Ableton Live) all use representations (either 32-bit float or 32-bit PCM) significantly "larger" than the 24-bit integers that most audio interfaces use. It seems worthwhile to simply maintain that representation if possible all the way through the mixer to the final DAC before an analog gain stage at the final output and not waste time or incorporate flaws from DA/AD conversion if it's not necessary.
Is that even possible? Projects like soundflower and jack exist, so that leads me to believe that it is possible, but I have no idea how to do the hardware side even if I did manage to wrap my head around the code.
Thanks.
i dont think you realize how much work is involved, being vaguely aware of how much work is involved in the dsp side is not aware enough, given the complexity of your task. dsp will be where most of the work is done. i think the best chips for you task are going to be the crystal CS4382A, or ESS ES9018. however i think you would probably do better to just get a nice audio card that allows hardware acceleration of the software mixer and dsp functions. or just have a fast computer that can do all of this internally and output to a regular dac, while using the midi control surfaces i mentioned
i dont like your chances of diying anything even close to this novation twitch, or nocturn or one of the other keybord/controllers which are pretty reasonably priced for what they are, plenty cheaper than what it will cost you without the benefit of their R&D over the last few years working on exactly this type of setup. so the twitch has a dac and headamp, but you could use any dac for output. all mixing tasks are then done in the digital domain in the mac (assuming you have a mac based on your post) for a higher end output, not sure the twitch has digital output itself to feed another dac directly if you feel the dac is not up to the quality you want, or use nocturn as a pure controller and use whatever dac you want.
it seems that you dont realize that the reason the files are 32bit, is not because its using 32bit music data for each channel, but because its a multichannel format that includes the 24 bit audio data, the mixer channel strip positions, plugin data and positions, timeline edits, eq data and volume data. not because its higher quality.
this doesnt bode well for your ability and knowledge in the area to get this done.
with that in mind, buy yourself an interface with the I/O you need and add a control surface like the korg or novation units. DIY like this does NOT save money, it allows you to get exactly what you want, but in this case i just feel that they have already thought of all you want and more. i've got an SLkII controller and the feel is quite nice and rugged, i also use a korg hardware synth module so its more up my alley
if you really want to continue, you really need to do a whole lot more reading before you can ask meaningful, less general questions so that we can help you, people far more technically savvy with dsp design than i am NOT
i dont like your chances of diying anything even close to this novation twitch, or nocturn or one of the other keybord/controllers which are pretty reasonably priced for what they are, plenty cheaper than what it will cost you without the benefit of their R&D over the last few years working on exactly this type of setup. so the twitch has a dac and headamp, but you could use any dac for output. all mixing tasks are then done in the digital domain in the mac (assuming you have a mac based on your post) for a higher end output, not sure the twitch has digital output itself to feed another dac directly if you feel the dac is not up to the quality you want, or use nocturn as a pure controller and use whatever dac you want.
An externally hosted image should be here but it was not working when we last tested it.
it seems that you dont realize that the reason the files are 32bit, is not because its using 32bit music data for each channel, but because its a multichannel format that includes the 24 bit audio data, the mixer channel strip positions, plugin data and positions, timeline edits, eq data and volume data. not because its higher quality.
this doesnt bode well for your ability and knowledge in the area to get this done.
with that in mind, buy yourself an interface with the I/O you need and add a control surface like the korg or novation units. DIY like this does NOT save money, it allows you to get exactly what you want, but in this case i just feel that they have already thought of all you want and more. i've got an SLkII controller and the feel is quite nice and rugged, i also use a korg hardware synth module so its more up my alley
if you really want to continue, you really need to do a whole lot more reading before you can ask meaningful, less general questions so that we can help you, people far more technically savvy with dsp design than i am NOT
Last edited:
Thank you for your thoughtful reply.
Yeah…that is not what I want to do.
The Novation Twitch is a piece of crap designed for lousy software and based on a gimmick effect that I got sick of a long time ago. I'm actually pissed off about its existence because it means I'll have to listen to that effect more as every idiot who's ever seen DJTechTools.com mashes buttons randomly because he thinks doing so has something to do with mixing well. The other controllers you mentioned imply that you know less about DJing than you think I know about mixer design. They're all crap. They don't have enough controls, and they were designed by monkeys in board rooms. They're the kind of things that idiots buy when they realize–fifteen years late–that DJs have turned into rock stars and scream "Me too! But I don't want to spend a lot of money…or practice…or learn anything."
I own better midi controllers that have been relegated to piles.
If I wanted to use a midi controller as my mixer, I'd plug one in.
Next, I'm not talking about project files. And I do have a clue how representations of audio work in memory. You're just plain wrong about what you think is stored in these representations. I agree that it does not increase the quality of the source material, which are all 16-bit/44.1kHz wav files decoded from mp3s. But it also doesn't store anything about the rest of the environment. In software, you represent the same audio signal using a larger number space for a few specific purposes. In the DAW/production world, it's about maximizing quality for as long as you can. In the DJ world, it's mostly about avoiding clipping/compression to try and maintain the little dynamic range that's left in modern music.
Either it was unclear what I was referring to, or you know less about this topic than I do.
As for the chips your recommended. Thanks. They're not what I need. Obviously, I need a DAC at some point…but I'm just not worried about choosing the DAC at this stage. What I needed to know was whether it was possible to build a USB/FW interface that allowed me to stream the audio from software into the digital mixer without using a separate audio interface, not with building one. Apparently…yes, it is.
A 5 minute discussion with a hacker (with relevant experience) taught me that it is possible. Apparently, it's easier than I thought it was.
Analog Devices and TI both make components and prototype boards with that functionality. I'm sure others do as well. The prototype boards also tend to have DACs that are fine quality for the first stages of designing this thing.
What I need at this stage is to look through the documentation for different prototype boards to figure out how much I actually have to spend. I also need to spend some time in matlab designing my EQs, summing bus, signal routing, and a good limiter. That's going to be the hard part…not allowing clipping while avoiding compression as much as possible. It's definitely possible to do it, I just don't know all of the math yet.
I'm also not trying to save money compared to your conception of what DJ gear costs. I own and have given up on Vestax and NI controllers infinitely better suited to the task than anything Novation makes. The exact functionality I want is not possible with a controller except by using what I consider an unstable software setup. Plus, as much as I have embraced controllers for some parts of my gear, I like having a physical mixer that doesn't rely on a laptop. Obviously, I don't care if it's digital or analog, but I want it to be real.
I'm trying to build myself a mixer because it seems like an interesting project that would teach me a great deal and get me exactly what I want. The only DJ mixer on the market that appeals to me more than my A&H is the E&S DJR-400, a $2500 french boutique analog mixer that is missing exactly one feature that I consider essential. If I were willing to pay for it, I'm sure the builder would incorporate that feature…or I could do it this way and learn something in the process. If the total cost of the mixer I end up building is less than $2500 and takes less than a year, I'll consider it a bargain.
But, no, you're right…a $500 piece of crap made for software I don't like and less suited to my style of playing than the $1500 worth of controllers I have sitting next to my $1000 mixer and $800 audio interface is a good idea…or failing that, a $100 midi controller with a horrendous layout is absolutely a substitute for a $2500 boutique mixer that isn't quite good enough.
And I'll be sure to look into specific chips that have nothing to do with any of the questions I asked.
Since your reply was the only one in 4 days, either I have no idea how to solicit useful information from this board or there isn't any useful information on here. So, don't worry about how much research I'm doing. I have no confidence in your ability to help me.
Yeah…that is not what I want to do.
The Novation Twitch is a piece of crap designed for lousy software and based on a gimmick effect that I got sick of a long time ago. I'm actually pissed off about its existence because it means I'll have to listen to that effect more as every idiot who's ever seen DJTechTools.com mashes buttons randomly because he thinks doing so has something to do with mixing well. The other controllers you mentioned imply that you know less about DJing than you think I know about mixer design. They're all crap. They don't have enough controls, and they were designed by monkeys in board rooms. They're the kind of things that idiots buy when they realize–fifteen years late–that DJs have turned into rock stars and scream "Me too! But I don't want to spend a lot of money…or practice…or learn anything."
I own better midi controllers that have been relegated to piles.
If I wanted to use a midi controller as my mixer, I'd plug one in.
Next, I'm not talking about project files. And I do have a clue how representations of audio work in memory. You're just plain wrong about what you think is stored in these representations. I agree that it does not increase the quality of the source material, which are all 16-bit/44.1kHz wav files decoded from mp3s. But it also doesn't store anything about the rest of the environment. In software, you represent the same audio signal using a larger number space for a few specific purposes. In the DAW/production world, it's about maximizing quality for as long as you can. In the DJ world, it's mostly about avoiding clipping/compression to try and maintain the little dynamic range that's left in modern music.
Either it was unclear what I was referring to, or you know less about this topic than I do.
As for the chips your recommended. Thanks. They're not what I need. Obviously, I need a DAC at some point…but I'm just not worried about choosing the DAC at this stage. What I needed to know was whether it was possible to build a USB/FW interface that allowed me to stream the audio from software into the digital mixer without using a separate audio interface, not with building one. Apparently…yes, it is.
A 5 minute discussion with a hacker (with relevant experience) taught me that it is possible. Apparently, it's easier than I thought it was.
Analog Devices and TI both make components and prototype boards with that functionality. I'm sure others do as well. The prototype boards also tend to have DACs that are fine quality for the first stages of designing this thing.
What I need at this stage is to look through the documentation for different prototype boards to figure out how much I actually have to spend. I also need to spend some time in matlab designing my EQs, summing bus, signal routing, and a good limiter. That's going to be the hard part…not allowing clipping while avoiding compression as much as possible. It's definitely possible to do it, I just don't know all of the math yet.
I'm also not trying to save money compared to your conception of what DJ gear costs. I own and have given up on Vestax and NI controllers infinitely better suited to the task than anything Novation makes. The exact functionality I want is not possible with a controller except by using what I consider an unstable software setup. Plus, as much as I have embraced controllers for some parts of my gear, I like having a physical mixer that doesn't rely on a laptop. Obviously, I don't care if it's digital or analog, but I want it to be real.
I'm trying to build myself a mixer because it seems like an interesting project that would teach me a great deal and get me exactly what I want. The only DJ mixer on the market that appeals to me more than my A&H is the E&S DJR-400, a $2500 french boutique analog mixer that is missing exactly one feature that I consider essential. If I were willing to pay for it, I'm sure the builder would incorporate that feature…or I could do it this way and learn something in the process. If the total cost of the mixer I end up building is less than $2500 and takes less than a year, I'll consider it a bargain.
But, no, you're right…a $500 piece of crap made for software I don't like and less suited to my style of playing than the $1500 worth of controllers I have sitting next to my $1000 mixer and $800 audio interface is a good idea…or failing that, a $100 midi controller with a horrendous layout is absolutely a substitute for a $2500 boutique mixer that isn't quite good enough.
And I'll be sure to look into specific chips that have nothing to do with any of the questions I asked.
Since your reply was the only one in 4 days, either I have no idea how to solicit useful information from this board or there isn't any useful information on here. So, don't worry about how much research I'm doing. I have no confidence in your ability to help me.
off you go then, i recommended something that didnt cost stupid money, of course there is more (the korgs for example), but im not about to recommend someone on a diy board spend 10k. and you only have to look at what is available to know you can stream multichannel, or any data both ways over usb/fw, both of which are dead technologies imo.
but if you want to beat what is here already, it will take you years to do all the dsp programming with something like a xilinx chip, which is where its all going to be done if you want in hardware, or years of software programming if you dont. there is nothing higher available in dacs and adc audio chips generally that is higher than 32bit, anything higher is done with dsp, as if you use the 32bits for audio, you dont have anything left for the functionality you want.
designing to cope with the crap idiot producers crank out sounds like fun, surely you only need the headroom in the mixer to cope with layering channels and making it too hot, not the channels themselves, or are they really that dense?
also, you might want to post this in the right forum, for the heart of your problem you want digital source, not line level
but if you want to beat what is here already, it will take you years to do all the dsp programming with something like a xilinx chip, which is where its all going to be done if you want in hardware, or years of software programming if you dont. there is nothing higher available in dacs and adc audio chips generally that is higher than 32bit, anything higher is done with dsp, as if you use the 32bits for audio, you dont have anything left for the functionality you want.
designing to cope with the crap idiot producers crank out sounds like fun, surely you only need the headroom in the mixer to cope with layering channels and making it too hot, not the channels themselves, or are they really that dense?
also, you might want to post this in the right forum, for the heart of your problem you want digital source, not line level
Last edited:
xilinx, my bad. just about any project i have seen that needs dsp for audio lately is using the xilinx range of dsps. you will still need a multichannel dac for output and the 2 i mentioned; and mentioned for that purpose if you actually read what i wrote, are at the top of the heap as far as being able to talk to dsps for level and other control of the channels. if you made a sabre es9018 based mixer you would have one of the best sounding things at any show on the planet provided you can actually use it to good effect and arent just a tweaker. wonderful sounding dacs
noone else replied eh? but noooo it wouldnt have anything to do with the way the question was asked. there are people here that can and have done everything you want to do, but you have to make it easier for them to answer in what is probably limited time they have to spend on audio forums.
and no i havent read all of that wall of text/rant above
noone else replied eh? but noooo it wouldnt have anything to do with the way the question was asked. there are people here that can and have done everything you want to do, but you have to make it easier for them to answer in what is probably limited time they have to spend on audio forums.
and no i havent read all of that wall of text/rant above
Last edited:
- Status
- Not open for further replies.