7

I have a question about what parameters (such as bitrate) an audio codec would need to use to be psychoacoustically indistinguishable from stereo 10-bit PCM at a sample rate of 32,768 Hz.

Where would I ask such a question? It seems too basic of a question for Sound Design.

Improve this question
17
  • Without knowing anything about that subject, did you check dsp: dsp.stackexchange.com and vp: video.stackexchange.com/help/on-topic ? You know the drill, right? Check their help and ask on their meta before posting on their main site.
    – rene
    Sep 9 at 6:39
  • 2
    @rene It's certainly not Video.SE, but DSP.SE might be right, if they allow very basic question! I'll check it out.
    – forest distrusts StackExchange
    Sep 9 at 6:44
  • 1
    Looking through their questions, it appears my question would not be on-topic there. They seem to be dealing heavily with the theory behind the algorithms involved in signal processing, so while a question about the effect of MDTC in vorbis may be on-topic, "what bitrate should I use with vorbis to achieve transparency" would not be.
    – forest distrusts StackExchange
    Sep 9 at 6:56
  • 2
    [For if this makes it to another stack] 10-bit 32k PCM is already your 'codec'. PCM doesn't have a bit-rate, only a bit-depth. It is also already below the requirement to be 'psychoacoustically transparent'. That would generally be considered to be the adopted CD standard, 16-bit, 44.1kHz. You'd probably be OK on Sound Design with this, as these days far from being a hangout for professionals, it's where questions on how to connect basic equipment & 'why my mic buzzes' go to die.
    – Tetsujin
    Sep 9 at 11:32
  • @Tetsujin Is there a better term than transparent to refer to indistinguishable from lossless at a particular bit depth and sample rate? 10-bit 32,768 Hz PCM is what the hardware is capable of outputting, so using, say, 256 kbps vorbis would be overkill.
    – forest distrusts StackExchange
    Sep 9 at 11:34
  • Starting from such a crunchy sound anyway, you'd probably have to just test different algorithms & rates. The usual 'blurring' that happens in compressed audio really needs a higher frequency spectrum to do its 'tricks' in. I can't give you any practical advice as I haven't used anything below 16-bit 44k in nearly 40 years.
    – Tetsujin
    Sep 9 at 11:37
  • 2
    "It seems too basic of a question for" - this should never be a reason to look for a different network site, except possibly considering the Math Overflow / Math.SE split. The problem with beginner questions isn't that beginners need them answered; it's that beginners ask them (and thus communicate in the way that beginners do). The most important and valuable questions on Stack Overflow, as demonstrated by multiple objective metrics, are about fundamental tasks that any remotely experienced developer can accomplish without a second thought.
    – Karl Knechtel
    Sep 10 at 1:18
  • "Is there a better term than transparent to refer to indistinguishable from lossless at a particular bit depth and sample rate?" What's wrong with "indistinguishable from lossless"?
    – Karl Knechtel
    Sep 10 at 1:19
  • 1
    For what it's worth, Music: Practice & Theory takes quite a few questions about the theory of sound on a pretty technical level - as is necessary to understand the ideas behind microtonal composition, historical tunings and temperaments, etc. You might be able to get some interest there.
    – Karl Knechtel
    Sep 10 at 1:21
  • 2
    I would be careful, though, to phrase the question in a way that it can be answered objectively and not just with personal opinion. Whether or not there are audible differences between different audio codecs with different settings is a topic audiophiles love to have opinionated arguments about. Which often escalate into fanboyism for audio equipment and name-calling ("of course you can't hear a difference when you are using your cheap [shit-brand] headphones for under $1000").
    – Philipp
    Sep 15 at 8:32
  • @Philipp That is why I use the term psychoacoustically indistinguishable rather than "better", since that has a specific meaning and excludes placebo "improvements".
    – forest distrusts StackExchange
    Sep 15 at 9:36
  • You're asking a electrical engineering question (for which there's an entire sub-specialty within electrical engineering wrt. audio electronics design; as in, that specialty is the entirety of a substantial number of electrical engineering careers and many consulting firms that do just audio hardware design), so I would have expected Electrical Engineering to be the place to ask. However, your question is unclear as to what you're really wanting, as it's both naive and somewhat contradictory. It can also be viewed as too broad, as entire books have been written about such things.
    – Makyen
    Sep 15 at 13:51
  • 1
    @Makyen The question has nothing to do with electrical engineering nor audio hardware design. The sample rate and bit depth are already fixed in hardware and I'm neither changing that nor designing around it. I fail to see how the question is contradictory, but if the answer is "you must ABX test it and there are no heuristics you can use to make an a priori best guess", then so be it.
    – forest distrusts StackExchange
    Sep 22 at 19:16
  • @forestdistrustsStackExchange No, as currently written, the question is, definitely, electrical engineering. If you want it to not be about electrical engineering, then the question needs to define the specific hardware about which you are talking. As it currently is, it's extremely open-ended and asking about any codec and any hardware and all parameters of that codec+hardware. All of which are relevant to the question as written. The fact that for a comprehensive answer multiple different codecs and codec technologies need to be discussed is part of what makes it so broad.
    – Makyen
    Sep 22 at 19:48
  • As to "psychoacoustically indistinguishable", determining that is quite a bit about opinion for most people. To actually get a decent idea as to the specific hardware/software/settings, You could buy/rent test equipment worth in the ballpark of $10k and perform a large number of audio quality tests to compare different hardware, or even just different settings. However, even then you'd be in the ballpark of if the differences were actually "psychoacoustically indistinguishable" or not, which would have at least some opinion to it, but could be based on professional experience.
    – Makyen
    Sep 22 at 19:53

0

Reset to default

You must log in to answer this question.

Browse other questions tagged .