Oversampling vs. Upsampling

I see conflicting decice specs using the term ‘upsampling’.

Do people agree with this defintion of the term?

I am most likely wrong herebut my conception of these words are different.
Oversampling is the process of “reading the same samples X number of times and then feeding the higher data rate stream into the DAC section”.
Upsampling however, i consider to be “mathematically interpolating and/or adding intermediate bits to a bit stream to insert data where there were none initially”.

Some more info, from the “source” you could say:

From a practical standpoint they are the same. There is no implication about whether the ratios of the two sample rates are integer multiples (power of two) or whether the process is carried out synchronously or asynchronously. In the broader sense of sampling theory, ‘oversampled’ refers to the idea that the sample rate of a digital signal is higher than it needs to be; that is, it is higher than twice the frequency of the highest frequency component of the signal. Upsampling may then refer to the actual filtering+interpolation process that generates a higher SR signal from the original. But oversampling is used commonly to mean the same process, since upsampling produces an oversampled signal.

Yes, I understand. It’s the ‘misleading’ language in some adverts that bothers me.

That is true, but the problem is that manufacturers don’ t use the term ‘oversampling’. They only say ‘upsampling’.

There are two distinct definitions that refer to two different algorithms and that shouldn’t be used interchangeably, IMHO. At the very least, it bugs the retentive part of my nature. :smirk:

Does it make a difference to the end product?
I guess not.

Thanks for the reply…

Even if will not solve your question/doubt I find this a good article: https://www.audioholics.com/audio-technologies/upsampling-vs-oversampling-for-digital-audio

Thanks for the article.

The take away, for me, is that upsampling refers to an ‘arbitrary rate’, i.e. not necessarily a power of 2.

Yes, I would say. But also, historically oversampling was invented to allow real-time digital filtering with more than first order filters. For me this is the most important aspect. Then followed non-causal filters that compensate phase delays, all techniques that helped to make cheaper and more precise (not necessarily truth) audio systems.

1 Like