Announcement

Collapse
No announcement yet.

ALSA: Audio Compression Offloading, Power Savings

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by uid313 View Post
    Now when PulseAudio does it, its okay?
    It doesn't.

    Alternate sample rates

    A lot of common (desktop) hardware supports multiple sample rates. Interesting, among these, are 44100 Hz and 48000 Hz, since all common sample rates can be expressed as a simple multiple of one of these, which implies cheaper resampling, when it is required. Previous versions of PulseAudio only supported opening the device at a single sample rate, requiring all streams that did not match this rate to be resampled. We now support switching the device's sample rate dynamically at run time, allowing us to avoid resampling, or reduce resampling overhead. This should result in CPU and power savings on hardware that supports such switching (most Intel HDA-based devices at least).
    -Source

    Comment


    • #12
      Originally posted by Teho View Post
      It doesn't.

      -Source
      So I just play a music file and it will be bit-perfect?
      How do I do bit-perfect playback?

      Comment


      • #13
        as far as looking if your audio gets resampled, you can check that by checking the cpu usage of the program that the sound goes thru
        (some is used just to transfer the data, alot more is used for resampling; how much depends on your cpu ofc)
        idk much about pulseaudio, but common sense tells me you'd have to resample all audio to some common frequency and bit rate

        on most sound cards the signal gets resampled anyway

        as for the things written on boxes:

        dolby surround is a format for multi channel audio mostly used in dvd's
        eax is a proprietary protocol for 3D audio (yes, proper 3D audio is science and takes a little computing power, eax is an engine on creative sound blaster cards)
        most modern sound processors on cards support things like openal and others for 3D sound positioning, but im having my doubts as to how much is implemented on the software side in linux drivers/sound API's
        eax and things like that need to be supported in your game too, they rarely are

        most modern cards have mp3/flac/aac/etc decoders in their chips, but again they have to be written in the driver to be used
        only things you can read from a card and be sure it means something usefull is sampling rate/bit rate (but again you probably dont have music/games that go above 48khz/16bit so it dosent rly matter)
        and S/N ratio (signal to noise ratio, and i see lots of those data manipulated or just plain faked)

        most other things written are just bullshit

        interesting thing to know; best surround youl get is from headfones
        yes, i know it sounds dumb getting best surround from a stereo source, but when you look at it you have only 2 ears
        and the way we recognize where the sound is comming from is by its phase (and thats what 3D positioning means basicly)

        PS mp3/ogg/aac etc (and games use them or similar) are faaaaar from the original
        they use "psychoacoustics" too cut out a part of the sound spectrum that you probably wont notice missing
        that woudnt be bad at all if properly encoded, but almost all i found is just wrong either by the person encoding it(check "loudness wars" for an example) or by the fact that your sound source(speakers and all else in the chain, mostly speakers) isnt perfect

        for example you have a DnB song with vocals;
        when the bass gets loud, the mp3 encoder presumes you wont hear a part of the vocal
        and it would be right if your sound source could put out bass that loud(and in phase), but it cant
        so the vocal you can hear clearer then you should(since the bass aint that loud) you hear "zooming" in and out

        if you want a good sound card i recommend a cheap asus xonar (DS/DX, whatever is cheaper) since i found they have professional quality sound (i say professional quality sound, not that they are professional quality; but they still ALOT better then all in the price range)
        Last edited by gens; 08-30-2012, 09:07 AM.

        Comment


        • #14
          Originally posted by uid313 View Post
          So I just play a music file and it will be bit-perfect?
          How do I do bit-perfect playback?
          You set it up so whatever sample rate your music is in is specified as an alternate sample rate and make sure that the audio stops before something else with a different sample rate.


          Also, bit perfect isn't nearly as big a deal as people in this forum seem to think.. So long as you select a halfway decent sampler it won't degrade your audio perceptibly.

          Comment


          • #15
            Seams any topic can be made in to a discussion about PA. Also isn't all this answered in the manual, except for the stuff that is relevant for media players and not sound servers.

            Comment


            • #16
              Originally posted by gens View Post
              as far as looking if your audio gets resampled, you can check that by checking the cpu usage of the program that the sound goes thru
              (some is used just to transfer the data, alot more is used for resampling; how much depends on your cpu ofc)
              idk much about pulseaudio, but common sense tells me you'd have to resample all audio to some common frequency and bit rate

              on most sound cards the signal gets resampled anyway
              Your common sense has led you a stray. If i set my soundcard to use 96khz, then use jackd @ 96khz ( which in turn means every jack client/app will be using 96khz) then there isn't any resampling going on. However, if i were to use alsa-to-jack plugin or a2j-zita for youtube videos - then i would be resampling (since youtube videos aren't using 96khz, as far as i know).

              Originally posted by gens View Post
              and S/N ratio (signal to noise ratio, and i see lots of those data manipulated or just plain faked)
              S/N ratio is more important to proaudio than intelHDA consumer grade soundcard junk. I could see some of these companies faking numbers, but they may not be - since i have also seen quite a bit of variance in S/N ratios amongst even those cheaper cards. I am curious though, when making the claim that the data has been manipulated or faked ~ how did you determine this, exactly (?) and do you care to share your method + findings and companies at fault, please?

              Originally posted by gens View Post
              interesting thing to know; best surround youl get is from headfones
              yes, i know it sounds dumb getting best surround from a stereo source, but when you look at it you have only 2 ears
              and the way we recognize where the sound is comming from is by its phase (and thats what 3D positioning means basicly)
              it not only sounds dumb - but IS dumb. We may have 'only 2 ears', but those 2 ears do allow us to be able to tell what direction sounds are coming from with a fair amount of accuracy, in the real world. surround sound systems (5.1, 7.1, 10.2) provide a better 'surround sound' experience than head phones, no contest. Spend any time in a properly setup, high-end home cinema or movie theater and this becomes really obvious... The emulation used to make stereo into 3d/surround is okay, but doesn't touch anything physical, in the real world nor systems designed for surround.

              Originally posted by gens View Post
              PS mp3/ogg/aac etc (and games use them or similar) are faaaaar from the original
              they use "psychoacoustics" too cut out a part of the sound spectrum that you probably wont notice missing
              that woudnt be bad at all if properly encoded, but almost all i found is just wrong either by the person encoding it(check "loudness wars" for an example) or by the fact that your sound source(speakers and all else in the chain, mostly speakers) isnt perfect
              The loudness wars have very little to do with encoding, specifically (although, people encoding to mp3/ogg/etc do have a tendency to try to make recordings louder, compressed, eq'd, etc ~ which they shouldn't be doing in the first place, being as 99.999% of them have zero experience mastering audio). The loudness wars very specifically has to do with mastering techniques used in studios, since the rise of digital audio recording... For example, in wikipedia they show wavs of michael jackson's song 'black or white' ~ with every re-master / re-issue the recording gets louder ~ but this isn't because of the encoding, it is because of the re-mastering of the original recordings.

              http://en.wikipedia.org/wiki/Loudness_war

              As for ALSA - this all sounds like great news, and i am glad to see some improvements coming down the pipe, it can only mean good things. I'd like to find a little more information on 'wallclock timestamp support' and i am also interested in the proposed 'channel mapping API' ... hopefully, we will hear more about this soon.

              Comment


              • #17
                wait, means you only have 96khz music and you only play games that have 96khz sound etc...

                S/N ratio(and hum, that is directly related to it) is important for consumer grade too if you ask me
                for example i got big ass rly sensitive speakers that amplify the hum a lot, on the other hand bad amps put in small speakers amplify it a lot(not because they are better/stronger or whatever but because small speakers dont have a dynamic range to speak off so you turn them up a lot)

                S/N ratio is clearly faked on motherboard sound cards like your IntelHDA since in that case it depends on your power supply(ever heard wierd hissing when copyng a file?)
                on other cards i see it faked by shady test conditions like "96db SNR at 3khz and -9db" what is best for the card, but not standard
                all published data on hardware should be tested by a set standard, in case of audio either ISO or AES

                funny thing is that digital audio sources should have rly good figures, but assholes save money on simple filters(that go on the power supply of the opamp on the soundcard board)

                emulation ?
                our ears pick up a lot more dynamic range(they are so sensitive that 0dB is when its so quiet that you cant hear anything else cuz your heart is too loud) then our brain can process
                then our brain "calculates" the position of the source of sound based on phase difference(watch a documentary about sharks, its best explained there)

                20.4 surround or whatever figure they think up next to sell you crap cant work as good as headfones since your room acoustics will mess it all up
                (to set it up properly you need special speakers(i dont mean expencive, but rly special as in i seen one company that makes them and one school that made a few to teach students acoustics) also a lot of calculations and software control of phase, or you could just use good headphones that eliminate the room part out of the equation)

                to make it clear to you: "good headphones are A LOT better then the best speakers!" cut they sit directly on your ears, and they have rly light membranes(less sound coloration due to better control of the "driver"(the membrane+coil))

                as for the loudness war part, lots of people told me a song is crap encoded cuz its quiet
                "remastering" is usually done by experienced people that know what they do, and they do it for money because publishers want their songs to be louder then the competitions on some average hi-fi stereo system
                that has even more of an effect on encoding lossy audio like mp3 cuz of the psychoacoustic thing i explained earlier

                also your special 96khz software setup probably still resamples everything to 48khz in software since its probably what your card eats so using it too play youtube takes at least 2 resamplings, bout a bit loosy
                (not that you can hear the difference anyway)

                PS i like prodigy, but put a song from them in audacity

                Comment


                • #18
                  Originally posted by gens View Post
                  wait, means you only have 96khz music and you only play games that have 96khz sound etc...
                  no, in the example i gave (using jackd) both the card and all applications are using 96khz - no resampling. The next part of the example used alsa-to-jack plugin ~ which would resample. The same would be true of games, and music. But my 1st example was to show that you aren't always resampling, as you said.

                  Originally posted by gens View Post
                  S/N ratio(and hum, that is directly related to it) is important for consumer grade too if you ask me
                  for example i got big ass rly sensitive speakers that amplify the hum a lot, on the other hand bad amps put in small speakers amplify it a lot(not because they are better/stronger or whatever but because small speakers dont have a dynamic range to speak off so you turn them up a lot)
                  i agree it is important, but less important than what a proaudio / audiophile would want/need.

                  Originally posted by gens View Post
                  S/N ratio is clearly faked on motherboard sound cards like your IntelHDA since in that case it depends on your power supply(ever heard wierd hissing when copyng a file?)
                  on other cards i see it faked by shady test conditions like "96db SNR at 3khz and -9db" what is best for the card, but not standard
                  all published data on hardware should be tested by a set standard, in case of audio either ISO or AES
                  I only use professional grade audio interfaces (ie: i currently use RME and echoaudio)... IntelHDA (and onboard cards, in general) are crap - so i can't comment on noise you may hear using one. regardless, i was more interested in you actually showing the techniques, comparisons, testing you used to verify these vendors are FOS, but obviously you didn't actually do this.

                  Originally posted by gens View Post
                  emulation ?
                  yes, emulation. 3d sound via a stereo signal, as opposed to a room with multiple sound sources...

                  Originally posted by gens View Post
                  our ears pick up a lot more dynamic range(they are so sensitive that 0dB is when its so quiet that you cant hear anything else cuz your heart is too loud) then our brain can process
                  then our brain "calculates" the position of the source of sound based on phase difference(watch a documentary about sharks, its best explained there)
                  I don't need to, i already know this stuff. plus, there are better white papers, articles, etc floating around the web on the subject.


                  Originally posted by gens View Post
                  20.4 surround or whatever figure they think up next to sell you crap cant work as good as headfones since your room acoustics will mess it all up (to set it up properly you need special speakers(i dont mean expencive, but rly special as in i seen one company that makes them and one school that made a few to teach students acoustics) also a lot of calculations and software control of phase, or you could just use good headphones that eliminate the room part out of the equation)
                  I guess you've never heard of acoustic treatment before, eh? (which is common in any studio space, professional hi-end media room, etc). It is true that headphones will give you a 'room-less' listening experience - which can be helpful but ultimately headphones fall short, when it comes mixing or listening. ~ they are a good guide, but you will always have to listen to what you are mixing, eventually under proper studio monitors or speakers for the 'full experience' and full frequency range.

                  Originally posted by gens View Post
                  to make it clear to you: "good headphones are A LOT better then the best speakers!" cut they sit directly on your ears, and they have rly light membranes(less sound coloration due to better control of the "driver"(the membrane+coil))
                  To make it clear to you, I own $300+ headphones and $2000 studio monitors ~ and although i tend to mix using both, speakers offer something VERY important that headphones do NOT - headphones tend to limit or exaggerate the frequency range. ~ this is problematic, since you are limiting what you are actually hearing by wearing them. You miss out on so much of the body of the sound, for a slight trade-off/increase in detail... So honestly, i completely disagree with you here. Headphones are not a lot better then the best speakers....

                  Originally posted by gens View Post
                  as for the loudness war part, lots of people told me a song is crap encoded cuz its quiet
                  "remastering" is usually done by experienced people that know what they do, and they do it for money because publishers want their songs to be louder then the competitions on some average hi-fi stereo system
                  that has even more of an effect on encoding lossy audio like mp3 cuz of the psychoacoustic thing i explained earlier
                  I wasn't arguing that point. (i actually agree with you here). What i was arguing was that the 'loudness war' had anything to do with encoding (which AFAIK it doesn't), But came to be via professional audio engineers in studios, largely when digital audio recordings became popular (but even arguably before that, in the days of vinyl).

                  Originally posted by gens View Post
                  also your special 96khz software setup probably still resamples everything to 48khz in software since its probably what your card eats so using it too play youtube takes at least 2 resamplings, bout a bit loosy
                  (not that you can hear the difference anyway)
                  No, it does not resample to 48khz, i am not using some cheap sound card like you are probably used to using.
                  Last edited by ninez; 08-30-2012, 05:00 PM.

                  Comment


                  • #19
                    so your just telling me you have rly expensive audio equipment and therefore know better then anybody else ?

                    whitepapers like this one ?
                    http://www.fp3d.com/papers/WhyHeadphones.pdf

                    if you look at the specs, the xonar cards i recommended are better on paper then RME and such expensive cards
                    (the explanation is... electronics what i doubt you know anything about so i dont wanna confuse you)
                    (rightmark says they are better in tests too..)

                    check out Audio Power Amplifier Design Handbook by Douglas Self
                    the part named "Misinformation in Audio" will tell you about some nonsense you see people talk around

                    speakers i talked about are special by the fact they are full ranged single driver speakers shaped like a tear(or an egg i guess) to avoid coloring the sound by removing the bouncing of waves off the speakers rims

                    headfones also should have more dynamic range and also important hold the phase steady (as they have only one light driver(easier to control with less current, less mass) and not two or 3 that cant be perfectly phase balanced)

                    if you rly want to know i do have a shitty sound card, but i have also professional grade big ass concert speakers (they costed a lot, but since i made most the work myself, they still cost a lot(not that the price guarantees quality)) with rly high dynamic range and sensitivity(i can get you papers on them, not that you showed any papers that prove i sayd anything wrong)
                    i dont have money for a sound card, but il get a xonar DS for ~40$ that sounds better then your RME (not that anybody can notice the difference)

                    you have a studio or you just like black heavy stuff ?

                    PS Sennheiser http://www.sennheiserusa.com/around-...id-size_504765
                    freq response: 17-22,500 Hz not enough ?
                    mind also i think they test against AES standard, meaning 17-22,500Hz +- 3dB what rarely any monitor can do
                    Last edited by gens; 08-31-2012, 01:17 PM.

                    Comment


                    • #20
                      Originally posted by gens View Post
                      so your just telling me you have rly expensive audio equipment and therefore know better then anybody else ?
                      No, i did not say that and please don't put words in my mouth, completely misrepresenting my position. :\ i didn't write that since i own expensive equipment that i am somehow better than everyone else - so STFU with fallacious nonsense like that.

                      Originally posted by gens View Post
                      Originally posted by from very end of paper above
                      Combining headphones and a subwoofer for the ideal game audio system

                      The part of real world audio headphones can't deliver is the part which comes in through the skin as much as
                      through the ears- those energetic roars, rumbles, and booms that shake the furniture and give that certain
                      something to the disaster movies we all love.

                      These low frequencies aren't directional (the wavelengths are too big to interact with the head and ears) so the gamer can use headphones and a subwoofer together without hurting positional quality. This could be the ideal game audio system.
                      So basically, (you say) headphones deliver across the board and are better than any speaker, except (as i pointed out) for the fact that lower-end frequencies don't really come through them, and more importantly (which i will point out now) really low-frequencies are felt more than they are heard - and even this article says for the best experience use a sub-woofer with headphones...interesting.

                      Headphones give some extra detail (due to reasons you have pointed out), and they do solve the 'gap' problem between speakers, but that doesn't make them better. In a properly acoustically treated room, with a high-end media/theater/studio surround-sound setup - the experience is much richer, to continue to argue that headphones are better than any speaker is silly, when you consider everything that is involved. I personally find headphones to be a good reference / starting point, but a good set of studio monitors in a decent space is better at the end of the day, and you do miss out if only using headphones.

                      Originally posted by gens View Post
                      if you look at the specs, the xonar cards i recommended are better on paper then RME and such expensive cards
                      (the explanation is... electronics what i doubt you know anything about so i dont wanna confuse you)
                      (rightmark says they are better in tests too..)
                      if you really think xonar cards are really that good - you live in a fantasy world. they are pretty low end.

                      Originally posted by gens View Post
                      check out Audio Power Amplifier Design Handbook by Douglas Self
                      the part named "Misinformation in Audio" will tell you about some nonsense you see people talk around

                      speakers i talked about are special by the fact they are full ranged single driver speakers shaped like a tear(or an egg i guess) to avoid coloring the sound by removing the bouncing of waves off the speakers rims
                      nobody was arguing any point on the specific type of speakers you were talking about.

                      Originally posted by gens View Post
                      headfones also should have more dynamic range and also important hold the phase steady (as they have only one light driver(easier to control with less current, less mass) and not two or 3 that cant be perfectly phase balanced)
                      Except they don't have more frequency range and this is why in any studio, you will see expensive studio monitors being used for mixing. headphones are good to a point but limit much of the low frequency range and you tend to not lose much of the body of the sound, for a trade-off or slightly more detail... and we've already been over this.

                      Originally posted by gens View Post
                      if you rly want to know i do have a shitty sound card, but i have also professional grade big ass concert speakers (they costed a lot, but since i made most the work myself, they still cost a lot(not that the price guarantees quality)) with rly high dynamic range and sensitivity(i can get you papers on them, not that you showed any papers that prove i sayd anything wrong)
                      i dont have money for a sound card, but il get a xonar DS for ~40$ that sounds better then your RME (not that anybody can notice the difference)
                      Did i say you didn't own some big rig PA? (nope). Again, you live in a fantasy world to think xonar's are good soundcards (in particular you think they are amongst the best it would seem) - when in reality, they are one step up from onboard soundcards. and by the way here is the RME card that i own;

                      http://www.rme-audio.de/en_products_fireface_800.php#5

                      Please explain to me how ANY xonar actually touches this card (they don't), not in quality, performance or stability/reliability... not to mention you can't even use a xonar in a professional studio and certianly they don't touch RME in terms of quality or features.

                      As far as showing white papers, anyone with any background in music production knows that headphones are limiting ,they are good for basic mixing, but eventually you will want to be using proper speakers. So ya, you are wrong to think headphones are better than even the 'best' speakers.

                      Originally posted by gens View Post
                      you have a studio or you just like black heavy stuff ?
                      25+ years of being a musician and used to be an engineer - collected a lot of gear lots of studio equipment, rackmounts, instruments, etc.Other gear for both jamming/playing live and/or recording as well.

                      Originally posted by gens View Post
                      PS Sennheiser http://www.sennheiserusa.com/around-...id-size_504765
                      freq response: 17-22,500 Hz not enough ?
                      mind also i think they test against AES standard, meaning 17-22,500Hz +- 3dB what rarely any monitor can do
                      ..and yet, in any studio you will will probably be using much higher-end headphones than these (low-end pro grade) and will still want to use your studio monitors as headphones are not good enough, when you need to hear everything, in particular the low-end.

                      sorry but no dice and besides we are getting waaayy off topic, continuing is sort of pointless. You think headphones are better than any speaker/sound system, when in reality they don't deliver in the bottom-end and thus don't deliver those particular frequencies properly in the 3d space to be heard via those headphones that are better than any speaker.

                      I for one know that sound/music/etc is something that is not only heard but felt, to say that headphones are better than any speaker is ridiculous and it seems you are arguing this point just to argue.

                      Comment

                      Working...
                      X