Haha yeah I saw on the DM7 thread Charlie started that you weren't too impressed with it. Been looking at the Okto as well. Looks pretty sweet.. maybe someday
I use some software that I wrote. The best software DSP as of today is probably CamillaDSP, so I would advise looking into that.
BTW, the DM7 is best when used with balanced equipment (amplifiers). The performance is not as good if you adapt the TRS outputs to RCA. There is a thread on the DM7 in this forum as well as on audiosciencereveiw.com where you can get various opinions on this issue. I use balanced amps, since then the system is relatively immune from any interference noise, hum, etc which a computer can generate. This was more of a problem in the past when I was using the Raspberry Pi platform, but much less of a problem now that I use mini PCs using Intel or AMD processors.
I use a basic dbx and use the standard built-in autoEQ (single mic position). A friend has the posher model with the multiple mic positions! I have always wondered if multiple measurement mics in different positions summed to a single measurement input would have any merit.
Oh right. I think you mentioned you use your own software at the sdc. Cool stuff. Anyways, I'll check out CamillaDSP.
Noted, about the balanced amps. I hope to make something balanced at some point, but until then I do have an 8 channel Hypex UCD amp with RCA inputs. This will make noise without anything blowing up, yes? It just won't sound very good? Definitely wouldn't be a long term solution. Rather, a placeholder until I can afford to buy/build something balanced.
As for computer, is there anything else that you look for, other than Intel/amd based pc? I heard some people prefer fanless PC's? I didn't see the reason but I've read it mentioned
Fanless is nice, and if you stick to IIR filtering the CPU requirements are pretty low. You could check out the Mele Quieter 3Q and related models for example. For a step up in CPU power look at Ryzen 5 5500u and 5560u CPUs, for example in the Beelink SER 5 lineup (with fan, but a quiet one). I use Beelink U59 Pro with the N5105 (not the N5095) for a couple of systems, which have a fan but it is essentially silent.
If you will also be playing video (e.g. you want to also watch movies on the computer or whatever while audio processing) aim for the Ryzen 5 models.
If you will also be playing video (e.g. you want to also watch movies on the computer or whatever while audio processing) aim for the Ryzen 5 models.
I'm running everything through JRiver which has a 64 bit digital volume control. You can't get any cleaner than that, even the most expensive analog vol pot is left in the dust for noise etc.
Jan
I'm using DSP in a hybrid active/passive system. I'm probably doing not wrong, but it's working quite well for me just the same. I'm bi-amping a pair of speakers and using a Dayton Audio DSP-408 to split the signal to the mid and tweeter horns at 108.5db sensitivity, driven by a modded, 17wpc Glow Audio Amp Two tube amp and 99db sensitive 12" woofers via a 50WPC Emotiva Mini-X A100. The passives are actually doing most of the crossover work; the DSP is operating filters to reduce driver overlap with a steeper slope, as woofer extends well past the 400Hz crossover point, up to 4000Hz, iirc. I'm also using the PEQ in the DSP to increase the low end response- the woofer cabinet combination rolls off at 38Hz, but mild PEQ filters allow it to extend to 30Hz and negate the need for a subwoofer for music reproduction.
That you mention video suggests that your system is suitable for such playback? Is the latency from all the DSP processing small enough for video to play in sync with audio without any tweaking?
With IIR no issue.
With FIR, MANDATORY to have a video delay.
With FIR, MANDATORY to have a video delay.
Using CamillaDSP and a good CPU you can evidently get down to about 20msec delay, which would be imperceptible. You can look through the CamillaDSP thread for more info or just post a question about it there. I don't combine video watching and computer DSP, so I can't say for sure.
20ms is perceptible but if you loosely watch lipsync it can be acceptable.
It vary from people to people but i had a guitarist which was bothered at 5ms in direct monitoring of his own intrument...
20ms won't be very steep xover in low end at least... and forget about room correction.
It vary from people to people but i had a guitarist which was bothered at 5ms in direct monitoring of his own intrument...
20ms won't be very steep xover in low end at least... and forget about room correction.
The delay has nothing to do with the crossover or low frequencies. It is due to buffering within ALSA and the DSP processing software.
We are talking about IIR filtering only, so you can do lots of PEQ bands but not FIR filtering a la DIRAC or whatever room and impulse corrections where the kernel delay itself AND the processing time are much longer.
@krivium maybe you are thinking of group delay?
We are talking about IIR filtering only, so you can do lots of PEQ bands but not FIR filtering a la DIRAC or whatever room and impulse corrections where the kernel delay itself AND the processing time are much longer.
@krivium maybe you are thinking of group delay?
Last edited:
At some point the resolution will get less, e.g. towards a 16 bit, 24 bit or 32 bit DAC. And max 16-23 "real" bits, depending on DAC quality. Fusion has 24.8 bit internal processing. Initially I was listening at -40..-60 dB. That kills quite some bits. Now I shifted this with approximately 20 or 30 dB, given amp and resistor mods. But I want also to shift towards PC processing. I will use an AMD 5700G HTPC that has plenty of processing power for up-sampling and other DSP tasks.
looks like JRiver with Mitch's help, might have solved FIR latency, even for long filters.
https://yabb.jriver.com/interact/index.php?topic=136225.0
https://yabb.jriver.com/interact/index.php?topic=136225.0
Ouch! 20ms of buffering... this is slow. My dsp have something like 0,5ms ( don't remember exactly) from in to out at 96khz. My computer with Rme sound card might be in the 1,5ms range for same condition ( under windows). I'm virtually unlimited about IIR treatments in both case and this number are given with those engaged.
I'm not thinking about group delay, was talking about FIR where the lower you xover and the steepest the slope the more you need latency of treatments: in the Lake, for 48db/octave FIR under 100hz you need 25ms latency ( or 50ms i can't remember at the moment, it's not specified in Tap number but in slope steepness in the gui).
Mark100,
At least they decide to implement latency compensation! It's great for video which was painful to use with FIR in previous JRIVER version. It could even make real time monitoring of video+audio possible ( with a global delay but not different that how 'live' materials are broadcasted for some years anyway!).
Last edited:
It seems to delay the video by the delay of the FIR kernel and processing. A neat trick, but what if your video feed cannot be delayed because it is live and not file playback?
Theorically it doesn't mater: image is buffered (delayed) by the value needed for the audio treatment. So you have a global latency equal to the value needed for audio treatments.
It is implemented in every DAW since circa 2005. Before we had surprise with mixes involving plug ins as they were not time aligned to a common reference and has they did not require same duration of treatments ( reverbs where particularly slow)...
As long as you don't monitor musicians real time ( who would do that with Jriver?) It won't be an issue.
It is implemented in every DAW since circa 2005. Before we had surprise with mixes involving plug ins as they were not time aligned to a common reference and has they did not require same duration of treatments ( reverbs where particularly slow)...
As long as you don't monitor musicians real time ( who would do that with Jriver?) It won't be an issue.
You don't seem to understand my point. Not every video source can be delayed. For example I get my movies and TV/video through a ROKU TV. The ROKU tuner is inside the TV and there is no way to re-route the signal before it hits the display and there is no video delay capability. There are digital and analog audio outputs that I can use as input to a crossover/processor but the video cannot be delayed. So this approach is a non-starter for me. The same might be said for cable and steraming TV and video. Unless you can bring both the video and audio signals into the computer before displaying them, you cannot do anything to them (e.g. delay or process).
Honestly, for my TVs I just connect an amp and a pair of passive speakers to it and call it a day. There is no audiophile music coming out of the thing anyway. I have a dedicated audio only system for critical listening that is set up to make the most of the room where it is located (not where the TVs are in my home).
Honestly, for my TVs I just connect an amp and a pair of passive speakers to it and call it a day. There is no audiophile music coming out of the thing anyway. I have a dedicated audio only system for critical listening that is set up to make the most of the room where it is located (not where the TVs are in my home).
Charlie i see your point but wonder if you have ever used the software or one like it?
It seems not to me: for something to out Jriver it have to go through it. If it doesn't this is moot discussing what it can do or not. It is the same from anything played by it ( picture, radio, files, streaming service,tv, ...).
If you can't route something to it's input ( through virtual cable or hardware) it won't obviously work. If you use your tv's tuner i don't even see the point to discuss this as it's not designed to work that way ( using the tv as tuner/input).
From your description you are not the target for this kind of software and that's all, the TV in this kind of implementation is only a screen.
For the ones which use it, delay compensation is not a trick but a feature which was lacking for such a long time... it's a very good news they finally implemented it.
It seems not to me: for something to out Jriver it have to go through it. If it doesn't this is moot discussing what it can do or not. It is the same from anything played by it ( picture, radio, files, streaming service,tv, ...).
If you can't route something to it's input ( through virtual cable or hardware) it won't obviously work. If you use your tv's tuner i don't even see the point to discuss this as it's not designed to work that way ( using the tv as tuner/input).
From your description you are not the target for this kind of software and that's all, the TV in this kind of implementation is only a screen.
For the ones which use it, delay compensation is not a trick but a feature which was lacking for such a long time... it's a very good news they finally implemented it.
Last edited: