I work in audio and had a thesis in DSP, so I’ll try to explain this. It is an interesting idea, and in some cases could work, but wouldn’t be practically useful in most.
So there’s 2 types of audio encoding: Lossy and Lossless. All audio starts as lossless, and in many cases is converted to lossy to reduce the file size. The processing for this is NOT like compression, and is somewhat context aware in that it removes frequencies you wouldn’t hear because something else is more present and causing your ear not to really hear it (this is called masking).
If you were to upscale something that is lossless, it would probably work. Barring any inter sample peaks, you’d be inferring additional points in a waveform and that’s fine. They’re actually some audio plugins that do this as an intermediate step when processing a signal.
If you try to upscale something that is lossy, you can’t recreate what was removed, because there isn’t a way to infer that information anymore. It would be like if you were trying to upscale a photo but you’d already removed a dog that was somewhat obscured by a man’s hand. Even if you upscale the picture you can’t add the dog without somebody telling you that it was there before removal.
The other part of the equation is “why?”, and while I’m a bit of an audiophile and I have my collection of lossless audio, the limitations of the system are typically the human ear. CD quality, (16-bit at 44.1 Khz), is really all you’d ever need. Most people can’t hear above 20 kilohertz (if you’re over 18, you’re lucky if you even get close to that). In digital audio, you can reproduce any frequency in equal to or less than half of the sample rate. With 44.1Khz, that frequency is 20,050hz. If you want to go really crazy, DVD quality (24 bit at 48Khz). I consider anything about that nice to have from a archival and measurement standpoint, but there’s no point in terms of human listening.
I absolutely did! I was really hoping to teach a class about audio at a nearby university, but was told last minute that they can’t give it to me for bureaucratic reasons. It’s my dream to teach someday so I’m hopeful that I will find another University that will let me teach
Based on the little experience I have of the way you talk about audio, I can say without a doubt that whomever you teach will count themselves lucky. You’re passionate and you speak clearly, not over complicating things nor dumbing them down. I’m going to definitely keep an eye out in the audio topics for your posts as well as keep my fingers crossed that you another chance to teach!
I work in audio and had a thesis in DSP, so I’ll try to explain this. It is an interesting idea, and in some cases could work, but wouldn’t be practically useful in most.
So there’s 2 types of audio encoding: Lossy and Lossless. All audio starts as lossless, and in many cases is converted to lossy to reduce the file size. The processing for this is NOT like compression, and is somewhat context aware in that it removes frequencies you wouldn’t hear because something else is more present and causing your ear not to really hear it (this is called masking).
If you were to upscale something that is lossless, it would probably work. Barring any inter sample peaks, you’d be inferring additional points in a waveform and that’s fine. They’re actually some audio plugins that do this as an intermediate step when processing a signal.
If you try to upscale something that is lossy, you can’t recreate what was removed, because there isn’t a way to infer that information anymore. It would be like if you were trying to upscale a photo but you’d already removed a dog that was somewhat obscured by a man’s hand. Even if you upscale the picture you can’t add the dog without somebody telling you that it was there before removal.
The other part of the equation is “why?”, and while I’m a bit of an audiophile and I have my collection of lossless audio, the limitations of the system are typically the human ear. CD quality, (16-bit at 44.1 Khz), is really all you’d ever need. Most people can’t hear above 20 kilohertz (if you’re over 18, you’re lucky if you even get close to that). In digital audio, you can reproduce any frequency in equal to or less than half of the sample rate. With 44.1Khz, that frequency is 20,050hz. If you want to go really crazy, DVD quality (24 bit at 48Khz). I consider anything about that nice to have from a archival and measurement standpoint, but there’s no point in terms of human listening.
I feel like you enjoyed writing this and even if you didn’t, I enjoyed reading it. Thank you for taking the time and putting in the effort.
I absolutely did! I was really hoping to teach a class about audio at a nearby university, but was told last minute that they can’t give it to me for bureaucratic reasons. It’s my dream to teach someday so I’m hopeful that I will find another University that will let me teach
Based on the little experience I have of the way you talk about audio, I can say without a doubt that whomever you teach will count themselves lucky. You’re passionate and you speak clearly, not over complicating things nor dumbing them down. I’m going to definitely keep an eye out in the audio topics for your posts as well as keep my fingers crossed that you another chance to teach!