January 6, 2017 at 10:40 am #65049
Hi, I just discovered your DPS system today. I’m piping its output into SoundFlower to go over to Logic Pro X and Audition.
I notice that whenever I modify the volume level in DPS, the DPSReflector sets its own output level, and then again sets the output level of SoundFlower. So if I ask for a 3 dB reduction, I actually end up getting about 6 dB. So to get close to 3 dB I have to specify 1.5 dB in the DPSReflector.
That seems like a bit of an unintentional bug to me. You can watch this happen with the Mac Audio Midi Setup program, on the Audio Devices page.
Whatever your algorithm is doing, it is slightly nonlinear, so I can’t specify a 3 dB reduction and actually get exactly that on a music track. It might end up being a bit more, or a bit less, than what I asked for. C’est la vie… (it actually works deterministically on the EBU Reference Noise track)
So far, I’m finding it very interesting to listen to the difference. One must be very careful in such A/B comparisons to get the A and B levels matched, or else, whichever is louder will always sound a bit better.
I actually found your DPS because I was looking for a way to un-smash severely compressed recordings from the past. Your DPS was recommended over on GearSlutz. Indeed, in slight contradiction to what your waveform pics on your web pages show, DPS really does expand the dynamics very nicely. It also does, as you say, keep the important frequencies leveled out for loudness control.
– DMJanuary 7, 2017 at 12:55 am #65050Dan RiceKeymaster
This is interesting stuff! Obviously we developed DPS as a standalone consumer level audio optimization product. However, we’d like to chat with you about what your doing with it. It sounds pretty neat. So, we’d like to try get in touch with you after CES. Can you please email me directly at email@example.com and we’ll get you in touch with one of our engineers.
DPS SupportJanuary 9, 2017 at 10:55 pm #65056lhunathParticipant
As an aside, I also use SoundFlower, but my use case seems to suggest DPS isn’t actually doing its own volume control.
I have a multi-audio device set up to play on both built-in and a bluetooth sound card.
When I set the multi-audio device as output device, volume control does not work. OS X doesn’t even let me change it.
When I set DPS as the output device, and the multi-audio device as DPS’ output device, volume control _appears_ to work, in as far as, OS X lets me change it up and down and the DPS device’s volume changes, but the output volume doesn’t _actually_ change. When I turn the volume all the way down, the output _does_ mute.
When I set DPS as the output device, and SoundFlower as DPS’ output device, and I connect SoundFlower to my multi-output device, volume works just fine. When I increase volume on OS X, both the DPS and SoundFlower device’s volume change, and the resulting output volume goes up or down accordingly.
I cannot attest as to whether the volume change is accurate in dB or doubled as suggested by dbm, in my last scenario; but the fact that the output volume doesn’t change when SoundFlower is not in the mix suggests to me that DPS is not doing its own volume control (I wish it did!).
What it looks like to me, is that DPS doesn’t actually do any volume control and instead just forwards or syncs its own volume control on to its output device, and relies on that device effecting the volume.
So; from my place, I wish DPS could do its own volume control, the way SoundFlower obviously can, and it didn’t forward its volume requests on to its output device.January 10, 2017 at 9:42 pm #65066xaqmusicKeymaster
Regarding volume control issues, we will certainly look into this and try some other methods to mimic SoundFlower’s behavior. The core DPS algorithm now contains a volume control API so this may be feasible. However, we will be limited to the features offered by the CoreAudio API.
Your observations about DPS processing being non-liner are completely correct. However, unlike most compressors that “smash” the signal as you describe, DPS responds much differently and can actually restore dynamics where our ears hear them best.
You will notice DPS changes the level differently for each song or media type so you will have to adjust accordingly if you want to match perceived loudness of the source.
We have used DPS in the past to conform video sound tracks to ITSU loudness standards with excellent results. Let us know if this works for you.January 12, 2017 at 12:14 am #65068
Yes, I actually measured the envelope expansion from DPS on a strongly compressed iMusic track, and give an example of this useful behavior (IMHO).
Regarding volume control, my observations in Audio MIDI Setup tool shows that the gains are being split between the output of DPSReflector and SoundFlower such that each gets close to the square-root of the requested gain in amplitude, or about half each in dB space. Very peculiar.
I ended up just pushing both to zero dB attenuation, and control the volume at the source, e.g., iTunes. There is some attenuation of the source arising from DPS when its processing is bypassed. I haven’t measured it precisely yet, but it seems in the neighborhood of 3-5 dB.January 12, 2017 at 1:39 am #65069
I would be really interested to hear more about your calibration procedure. You offer quite a few presets with the advice that we should choose whichever sounds best. But when you guys are setting up a new device-specific profile, you must have some calibration target to work toward.
I’m using Sennheiser HD650’s through a MOTU UtraLight right now, and I have no idea where that falls in the range of “correct” presets. So for now, I just settled on New York headphones.January 12, 2017 at 7:34 pm #65070xaqmusicKeymaster
New York Headphones is a good choice for a generally good overall effect. Calibration is subjective when tuning for existing consumer devices. We take human factors such as loudness discomfort level (LDL), perceived changes in frequency response as volume increases and more. We use more objective measurements when working with embedded systems to ensure compliance with speaker and amplifier tolerances. These measurements are usually taken using Audio Precision tools and/or oscilloscope using sine waves.
We are enjoying your articles on dynamic range restoration! This has been a topic of our presentations to OEM customers and it is wonderful to see our users recognizing this benefit.
I am interested in creating a few profiles that are specifically designed for this purpose. If you are interested, please start a new thread about dynamic range restoration and we’ll get in the weeds on the subject.
- You must be logged in to reply to this topic.