Upsampled Files and Native Hi-Rez files: Same Quality?

m2tech logoMarco Manunta from M2tech reacts on a question from a customer.

A few days ago, a customer wrote me to ask my opinion on why he couldn’t hear any difference between an audio file produced at 44.1kHz/16bit and the same file upsampled to 192kHz/24bit. He thought that the upsampling process would improve the sound quality of the file.

While it could be marginally so when it comes to the way a DAC and its output filters work at different sampling rates, one thing must always be clear: upsampling doesn’t add information nor improve the inherent sound quality of a file.

Upsampling means changing the sampling frequency of a file to a higher one. Typical is going from 44.1kHz to 176.4kHz. A side effect of this process is that the resulting new samples fit on a larger word than the original ones, so 44.1/16 generally translates into 176.4/24.

Does this mean that we have more information in the upsampled file than in the original one? No. Let’s see why, discussing about bandwidth and resolution.

The original analog program has been heavily band-limited prior A/D conversion. Converting it @ 44.1kHz required band-limiting @ 20kHz or so. All information beyond that frequency was lost forever.

If we upsample, we just rearrange the original information to fit a denser sample sequence, but we don’t (and we can’t) add information beyond 20kHz. In un upsampled file with 176.4kHz sampling rate originated from a 44.1kHz file, the power spectrum of the signal falls to zero beyond 20kHz, even if 176.4kHz sampling would allow for information contents up to 8.2kHz.

Same thing for 44.1kHz file originated from hi-rez masters: they were actually downsampled, with a digital filtering process prior downsampling to cut all excess band away. Results are same: all information beyond 20kHz are lost.

And the resolution? It remains the same as the original 16bit file, at best. This is because after A/D conversion, which implies quantization, the original signal is superimposed a noise, called quantization noise, which cannot be separated anymore from the original signal. Thus, when we upsample, we apply the process to both signal and quantization noise, resulting in a 16bit resolution signal represented on 24 bits.

Summarizing, upsampling won’t improve the sound quality of the file, it will only allow the DAC to operate in more relaxed conditions with regards to bandwidth and resolution, probably leading to a marginal sound quality improvement. I hope I don’t disappoint anyone.

Over hermanvandendungen

Almost 40 years of being with and in the world of high-end audio equipment. As distributor of Krell, Sonus Faber, Cello, Conrad-Johnnson and many more, initiator of brands like Kiseki, Cogelco, AH!, PrimaLuna, Mystère and maybe more to follow.
Dit bericht werd geplaatst in M2Tech en getagged met , , , , , . Maak dit favoriet permalink.

Geef een reactie - Laisser un commentaire

Vul je gegevens in of klik op een icoon om in te loggen. logo

Je reageert onder je account. Log uit /  Bijwerken )

Facebook foto

Je reageert onder je Facebook account. Log uit /  Bijwerken )

Verbinden met %s

Deze site gebruikt Akismet om spam te bestrijden. Ontdek hoe de data van je reactie verwerkt wordt.