Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Understanding Upsampling (Read 1714 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Understanding Upsampling

I'm trying to get my head around upsampling. I understand upscaling of a video so that your DVDs don't too awful on your 4K TV. But upsampling? When I upsample an 44.1 kHz audio file to 88.2 kHz, what is in those additional data which the upsampling adds? Noise? Silence? Or does the algorithm really trying to figure out what could be in those frequencies for this audio but what happens to the original audio? won't be information below 44.1 kHz of the original be moved above 44.1 kHz? Could I upsample an 8 kHz VoIP audio to 32 kHz and actually have a perceived gain in quality even though the phone never recorded that sound frequencies above 4 kHz?

I tried searching this but I failed for some reason.

Re: Understanding Upsampling

Reply #1 –
To the best of my knowledge, it is about aliasing. Inherent to DA conversion is that you get a mirror of the audio signal and its starting point is 1/2 of the sample rate. To get rid of it you need a very sharp filter. If you oversample say 8 times, everything of course remains te same but 1/2 fs is now at 176. This allows the use a far less intrusive filters.
A better explanation can be found here:
https://www.audiosciencereview.com/forum/index.php?threads/aliasing-imaging-and-upsampling-a-real-example.35687/
https://www.audiosciencereview.com/forum/index.php?threads/interface-mystery.13115/#post-392905
TheWellTemperedComputer.com

Re: Understanding Upsampling

Reply #2 –
The difference between audio & video is that audio has a DAC (digital-to-analog converter).   The output from a DAC is "smooth" continuous analog (no samples in the analog).   Up-sampling doesn't make it any smoother and it doesn't automatically add any information.

There are some "exciter" or "enhancer" effects (derived from the Aphex Aural Exciter) that can create "missing" harmonics.   That might help if you have an 8kHz file.    There is a Harmonic Enhancer plug-in for Audacity.

Re: Understanding Upsampling

Reply #3 –
Quote from: tehabe on
When I upsample an 44.1 kHz audio file to 88.2 kHz, what is in those additional data which the upsampling adds?

When you output audio you are outputting a continuous waveform, so in so much as the upsampling is implemented correctly it should be the same things between the points before you upsampled.  Just now you'll have them on the digital side and not just at playback.

Quote from: tehabe on
Or does the algorithm really trying to figure out what could be in those frequencies for this audio but what happens to the original audio?

Yeah, that is exactly what it does.

Quote from: tehabe on
won't be information below 44.1 kHz of the original be moved above 44.1 kHz?

No.

Quote from: tehabe on
Could I upsample an 8 kHz VoIP audio to 32 kHz and actually have a perceived gain in quality even though the phone never recorded that sound frequencies above 4 kHz?

No.  In fact, if the resampling works correctly you should have the identical audio on playback. 

Re: Understanding Upsampling

Reply #4 –
There are Analog upsampling and Digital upsampling.

It's easier to analog upsample. Like ripping LP and upsampling to 24bit 192Hz.

Upsampling digital content can be challenging since it is all about 00111010. To achieve proper upsampling, the key lies in the effective use of filters.
EZ CD Audio Converter

Re: Understanding Upsampling

Reply #5 –
Quote from: tehabe on
I'm trying to get my head around upsampling. I understand upscaling of a video so that your DVDs don't too awful on your 4K TV. But upsampling? When I upsample an 44.1 kHz audio file to 88.2 kHz, what is in those additional data which the upsampling adds? Noise? Silence? Or does the algorithm really trying to figure out what could be in those frequencies for this audio but what happens to the original audio? won't be information below 44.1 kHz of the original be moved above 44.1 kHz? Could I upsample an 8 kHz VoIP audio to 32 kHz and actually have a perceived gain in quality even though the phone never recorded that sound frequencies above 4 kHz?
Before talking about quality related topics, playing SD content on 4k TV without upsampling will result in a very small output video size. As for audio, the majority of hardware accept a minimum sample rate of 44.1 or 48kHz, without upsampling 8kHz input will be played in incorrect speed and pitch.

So if SD content fills the whole screen or fits the whole screen with an appropriate aspect ratio it is already upsampled. 8kHz audio played with correct pitch and speed is also upsampled.

For video the upsampling can be performed by the playback software, or a separate video renderer (e.g. madVR), or the GPU driver, or the TV. For audio it can be the playback software, a plugin, the OS mixer or the audio device driver.

Quote from: Roseval on
https://www.audiosciencereview.com/forum/index.php?threads/interface-mystery.13115/#post-392905
This post was to illustrate ADC aliasing caused by downsampling, not imaging caused by upsampling, but both can happen when using non-integer resampling ratios.