6 Replies Latest reply on Aug 31, 2010 6:29 AM by lundaudio

    Please correct sprctral aliasing.

      Both AA V.2.0 and the new AA V3.0 exhibit spectral aliasing on play. PLEASE FIX IT!!!!!!!!!!

      This did not occur in V1.5 and earlier, and also did not appear in any versions of Cool Edit. Thus showing the original authors at Syntrillium knew a whole lot more about practical signal processing than the Adobe people.

      To demonstrate this, create a low sample rate file - say 8kHz - and generate a swept sine from about 20Hz to 4khz. On playback, in addition to the generated swept tone tone you can also hear an HF tone sweeping downwards. This should not be there,and is not audible in earlier AA versions.
        • 1. Re: Please correct sprctral aliasing.
          SteveG(AudioMasters) Level 6
          >Thus showing the original authors at Syntrillium knew a whole lot more about practical signal processing than the Adobe people.

          ???? - they are the same people!

          >On playback, in addition to the generated swept tone tone you can also hear an HF tone sweeping downwards. This should not be there,and is not audible in earlier AA versions.

          I tried exactly what you suggested, and on this system I neither hear an additional tone, nor can see one on the spectral analysis. One of the significant differences between AA2.0 and 3.0 and earlier versions is that they use an ASIO driver (or the Audition Windows Driver, which is a converter for the Audition ASIO output), whereas 1.5 and everything before uses the native Windows audio system. The chances are that it is variations in how ASIO is being handled by different soundcards that is what is behind what you have experienced, and not explicitly Audition at all. There is absolutely no comparison between these two systems - the ASIO system completely bypasses a lot of the Windows OS, which means that it has lower latency, but unfortunately not all soundcard ASIO drivers are that good, it would seem - there have been a lot of issues with a lot of ASIO drivers, and a lot of people find that they get on better with the free ASIO4ALL WDM wrapper than their card's native driver.

          Could a dodgy driver cause aliasing problems? It most certainly could.
          • 2. Re: Please correct sprctral aliasing.
            Level 1
            Thanks Steve. I'd realized it was ASIO related - or at least that change coincided with the start of the problem. Your analysis may be right. However, since it occurs in ALL the PCs i've tried it on - several varieties of DELL & HP desktops and an ACER laptop, using various standard windows sound cards. I'm tending to assume it's not simply a isolated dodgy driver. I've also talked a guy on the Adobe helpdesk through how to re-create the problem and he agreed there was a problem. You can also hear aliasing in some Flash audio too.

            You won't see the aliased component in the spectral display though. It's purely a playback effect and if you resample from 8k up to 48k the problem goes away - as you'd expect.

            Incidentally, i wonder how many of the original Syntrillium people are still with Adobe?
            • 3. Re: Please correct sprctral aliasing.
              MusicConductor Level 3
              >Incidentally, i wonder how many of the original Syntrillium people are still with Adobe?

              Enough to create a strong continuity. The original creator is still on board.

              >You can also hear aliasing in some Flash audio too.

              I have rarely
              i not
              heard aliasing in Flash audio. Of all major players in web streaming, it has by far the poorest audio.

              The real issue here is that Audition handles the data accurately, but the ASIO-to-Windows real world playback is dependant on a weak link that may or may not be Adobe-related, but could benefit from some attention.

              At least in the meantime, resampling the 8KHz file (if for no other reason but to just listen to it once) to a normal rate will prove that Audition's handling of the data is fine.
              • 4. Re: Please correct sprctral aliasing.
                Level 1
                Not all serious audio work takes place at 44.1kHz and higher! To some of us in the telecoms world, 8kHz IS a 'normal' rate!!! As are 16kHz and 32Khz. We actually first noticed the aliasing problem whilst analysing the performance of some low bitrate telephony speech coders. Some differences in our opinions turned out to be due to me listening on AA2.0 and a colleague using AA1.5.

                Sadly, it's a bit of a drag to have to convert any telephony quality audio file to a higher sample rate before listening.
                • 5. Re: Please correct sprctral aliasing.
                  MusicConductor Level 3
                  That is certainly a legitimate feature request! Telephony-related questions have been posted here for years, so you're definitely not alone in needing low sample rates to play correctly. However, since the driver is ultimately what is responsible for the bad upsampling, the only part of this that Adobe can fix is to work with it in the "Audition Windows Sound" module, which would only benefit users who
                  i don't
                  already have an ASIO-compatible sound card.

                  Either that, or Audition would need a output resampling option that happens before the driver is accessed, e.g. all audio is sent to the driver at 48Khz regardless of how the hardware reports its abilities and the session's sample rate.
                  • 6. Re: Please correct sprctral aliasing.
                    lundaudio

                    Still hear this spectral aliasing problem when playing

                    8 kHz files from Audition 3.0, very noticable and makes it difficult to use the program for serious telecom work. Tried several sound cards with ASIO drivers. Tried also ASIO4ALL driver. Regardless of where the problem is, it would be great if there is a solution described somewhere, perhaps there already is? Thanks!