|
Post by goodrevrnd on Oct 11, 2016 18:02:26 GMT -5
I'm using GMMP on a Nexus in my car. My Nexus outputs to a Behringer UCA202 -> Optical Out -> JL Twk DSP -> Amp. While the UCA202 is a DAC, in this case it really only serves as a bridge from USB to optical and the TwK handles DSP/DAC duties. Given this setup, is the "Use External DSP" the best option to use in GMMP? Does this fully bypass all Android and GMMP audio processing?
Now for the curve ball. I use ReplayGain in GMMP. GMMP is really just for albums I've downloaded. I still use Spotify for most listening. The problem of course is that Spotify is significantly louder than GMMP due to ReplayGain (plus I'm convinced they've increased the levels on all their rips). This makes tuning the gain on my system a bit of a pain. I intend to solve this by adding preamp within GMMP, but before doing so I'd like to better understand the consequences of using GMMPs internal processing vs. the "Use External DSP" option, and also verify there is no way to add any preamp when "Use External DSP" is checked.
|
|
|
Post by GoneMAD on Oct 11, 2016 19:43:25 GMT -5
external dsp just disables gmmp's eq and lets an external application do any dsp effects. The decoding is still done by gmmp. Replaygain WILL apply regardless of the external dsp setting, however the preamp will not
As for your best option? I have no clue about any of that hardware so i cannot say. If you have a good hardware dsp i'd probably trust that over any software. GMMP's dsp is pretty good but its in software so it does consume more battery than if its off
|
|
|
Post by goodrevrnd on Oct 11, 2016 21:18:49 GMT -5
So it's really more intended for using another software DSP like viper? Is there any effective difference between disabling it and leaving it on with no settings changed other that preamp?
|
|
|
Post by goodrevrnd on Oct 11, 2016 21:22:15 GMT -5
Also if replay gain applies regardless, I'd propose a replay gain preamp level feature like power amp has.
|
|
|
Post by GoneMAD on Oct 11, 2016 21:43:26 GMT -5
So it's really more intended for using another software DSP like viper? Is there any effective difference between disabling it and leaving it on with no settings changed other that preamp? yes.. the external dsp setting will actually have gmmp broadcast an audiosession id out to other apps / dsps.. so they can attach to teh audio stream and apply effects. If your extra processing is in hardware there is no benefit to using it. Preamp is actually before it hits the dsp entirely so it will cause distortion if boosted too high (its been a long standing "bug" and on my todo list to fix but requires a signficant rewrite to portion of the audioengine). There is a default replaygain option tho that lets you set the default RG value if its blank. The preamp is your safer bet if you want to boost gain since the limiter will be applied to prevent clipping I am sure you are wondering why it works that way and its simply because i had not originally designed gmmp's audioengine for all those capabilities, so there were limitations when i started adding some of the new audio features.. nothing short of a redesign/refactor can fix it.. which i plan on doing at some point, but dont know when i'll get the time
|
|
|
Post by goodrevrnd on Oct 12, 2016 0:10:16 GMT -5
I RG tag everything down to 89db target level so in my use case the preamp is really more intended to attempt to get relative volume more inline with Spotify's horrendously loud music. Obviously I could normalize to a higher target level but that doesn't make sense for the rest of my audio equipment.
I'm not sure how your description of the preamp implementation is really a bug, but I'm no audio expert and will take your word for it, and thanks for the warning. You've been very helpful sorting all this out.
|
|
|
Post by GoneMAD on Oct 13, 2016 5:53:50 GMT -5
The bug is the replaygain being applied early before it ever hits the dsp.. Decoders = 16bit, dsp = 32bit. If gain is boosted when its 16bit it can basically hit the max value and gmmp has to cap it.. so you can get a lot of distortion. If your RG is a negative value to lower the total db, everything is all good and it sounds like thats the case for you
Signal-to-noise ratio and resolution of bit depths # bits SNR 4 24.08 dB 8 48.16 dB 11 66.22 dB 16 96.33 dB 20 120.41 dB 24 144.49 dB 32 192.66 dB 48 288.99 dB 64 385.32 dB
|
|