foo_dsp_effect, A effect array for FB2K Options

[mudlord]

Again, I failed to see the logic in adding said features.

[markanini]

Please consider effect for adding a small delay in either channel controlled by a custom tag field. I often find commercial releases with 10-20 ms delay between L-R.

[EpicForever]

Again, I failed to see the logic in adding said features.

Hmmm... in post #135 you said that you can add bypassing when effect value is set to "0%". You wondered about DUI elements ncessity, but bypassing was accepted proposition. I will be pleased even if you could add only bypassing at 0% especially in Playback Rate Shift effect . Without DUI/CUI elements . I already explained why bypassing at 0% matters - in this case there's no unnecessary filtering and resampling added to output when effect is not used (set to 0%).

[mudlord]

Please consider effect for adding a small delay in either channel controlled by a custom tag field. I often find commercial releases with 10-20 ms delay between L-R.

You mean differing behaviour IF such tags exists, and if not, acts as normal?

[markanini]

Please consider effect for adding a small delay in either channel controlled by a custom tag field. I often find commercial releases with 10-20 ms delay between L-R.

Oops, the measrements I was reading were reported in μm, the delay in ms is more like 0.05 -0.1 ms. Still it's enough delay to skew the apparent stereo field.

[EpicForever]

This delay between channels is added intentionally - in purpose of widening stereo basis. Stereo widening effects (like making stereo from mono signal) are always based on adding interchannel delay. So such DSP will be destroying effect rather than sound tuning effect.

[lvqcl]

Please consider effect for adding a small delay in either channel controlled by a custom tag field. I often find commercial releases with 10-20 ms delay between L-R.

It is possible to do this with foo_dsp_yac (but IIRC it works only for stereo: no multichannel support).

[markanini]

This delay between channels is added intentionally - in purpose of widening stereo basis. Stereo widening effects (like making stereo from mono signal) are always based on adding interchannel delay. So such DSP will be destroying effect rather than sound tuning effect.

I'm talking about real stereo recordings with a consistent inter-channel delay, possibly due to a tape transport issue. One such example is Yes - Close to The Edge by Rhino.

Please consider effect for adding a small delay in either channel controlled by a custom tag field. I often find commercial releases with 10-20 ms delay between L-R.

It is possible to do this with foo_dsp_yac (but IIRC it works only for stereo: no multichannel support).

I could work with that, thanks!

[EpicForever]

I downloaded latest version of "foo_dsp_effect" but bypassing at "0%" is still missing... It's a matter of sound quality, not just my whims...

[EpicForever]

OK, in newest version (released on 2012-06-25) there is bypassing implemented and it works properly Many thanks !!

[Fandango]

I really like the PRE_EMPHASIS feature, but there was one thing that was bothering me and it needed testing on my behalf: if this tag field is present and set to true, will it be in effect when converting the audio files?

My results are pleasing: when converting to a lossy format, it de-emphasis will be applied. When converting to a lossless format, it will not! And when using the dynamic range meter plugin, it again will be applied. When bit-comparing identical files, with and without the tag field, they are recognised as identical. Which all in all is exactly how it should be.

[kode54]

If it's a DSP filter, do you have it in your chain? If not, do you have additional processing enabled when converting?

[CoRoNe]

I don't know anything about foo_dsp_effect its inner workings, but why does Pitch Shift (and perhaps other effect-filters too, I don't know) change the duration?

Original test sample: 25.500s / 1124550 samples
SoundTouch plugin: 25.488s / 1124032 samples
Effects DSP (Pitch Shift) plugin: 25.479s / 1123640 samples
Avisynth (TimeStretch): 25.500s / 1124550 samples(!)

For everyday listening it's no problem, but for encoding I obviously prefer Avisynth at the moment.

[Fandango]

If it's a DSP filter, do you have it in your chain? If not, do you have additional processing enabled when converting?

No and no. It's not in the DSP chain and there's no need to enable Processing when converting. So I don't need an extra preset and I don't have to sort out pre-emphased records for converting seperately, which is great! It just works anyway just because of the tag field. :thumbs: All I have to take care of is adding the tag field when adding a record to my library.

Would be great if the plugin supported the "FLAG PRE" line in embedded cue sheets, not that I use embedded cue sheets anymore... but I think it would be a good addition to the plugin.


This post has been edited by Fandango: Aug 12 2012, 17:30

[mudlord]

there is no means atm to parse cuesheets, thats something Peter will need to add.

[mudlord]

I don't know anything about foo_dsp_effect its inner workings, but why does Pitch Shift (and perhaps other effect-filters too, I don't know) change the duration?

Original test sample: 25.500s / 1124550 samples
SoundTouch plugin: 25.488s / 1124032 samples
Effects DSP (Pitch Shift) plugin: 25.479s / 1123640 samples
Avisynth (TimeStretch): 25.500s / 1124550 samples(!)

For everyday listening it's no problem, but for encoding I obviously prefer Avisynth at the moment.

Hmm, AVISynth returns complete samples?
In the effect source for dsp_soundtouch.cpp, for get_latency, I return the amount of buffered samples divided by the samplerate to get current latency of the DSP. I run a buffer to basically get samples, and output them. I guess in the process there is a bug of sorts?

https://github.com/mudlord/foobar2000/blob/..._soundtouch.cpp

This post has been edited by mudlord: Aug 15 2012, 05:57

[EpicForever]

Hello again.

I'm not a programmer, so I kindly ask you mudlord - would it be complicated for you, to add one more feature to foo_dsp_effect library? I think about possibility of preserving information about preferred Playback Rate Shift, Pitch Shift and Tempo Shift for certain files in tags. It could be done like this: user manually adds tag field to his file - for exampla tag named "PRS" (for "Playback Rate Shift") and puts there a value - for example "+5" or "-10". Active DSP reads tag field and automatically applies shift. If values in fields will be greater than maximum values that are possible ("-150", "+300") for certain DSP, then the maximum value could be used - with optional warning from console. No autotagging - only manuall tagging done by user, then autoloading info. Optionally there could be additional check box in DSP configuration dialog: "automatically use shift values from tags". That would be perfect

Regards
Epic

[Rollin]

EpicForever
http://www.hydrogenaudio.org/forums/index....showtopic=96094

[EpicForever]

Hmmm.. It could solve my problem but to be honest I think it's too complicated or I'm just to lazy to configure DSP chains for every single track where I decide to change playback rate... At the moment it's more than 20 tracks. foo_dsp_effect has feature similar as requested in my previous post - setting up de-emphasis filter based on tags. So I think that implementation of my request shouldn't be a big problem.

[EpicForever]

Could I ask for any "developement news" for this component ? Is it posssible to add feature from my post : http://www.hydrogenaudio.org/forums/index....mp;#entry806411 ?

[mudlord]

current news: depends on my motivation to add features that I personally use. Since I don't use often features like this apart from de-emphasis, its very low on the current priority list.

[EpicForever]

Well, OK, but I understand that you have the procedure to read tags already coded. Only thing is to add it to other DSP and choose names for user tags... Numbers (values of tags) are just numbers, as the ones on fader. But I understand that you are doing it in your free time

[Eli]

Does foo_dsp_effect's de-emphasis recognize dBpoweram's pre-emphasis flag? If not, could this feature be added?

[Rollin]

Does foo_dsp_effect's de-emphasis recognize dBpoweram's pre-emphasis flag?

The tag "PRE_EMPHASIS" must be used on content which is supported by FB2Ks decode processor services, for this to work. Pre-emphasis can be signified in your media with the values "1" or "0".

[Eli]

I saw that but I'm not at home with my FLAC collection and don't know if this is the tag name dBpoweramp uses to flag pre-emphasis or not.