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: Increasing Sample Rates On An Audio Recording (Read 4701 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Increasing Sample Rates On An Audio Recording

I was recently told that if I increase the sample rate (from 44.1k to 88.2 or even 176.4k) of an analog audio recording that I'm digitizing, that when I later go to remove the noise inherent in vinyl, I can get a better result because I will be loosing less musical information when I try to get rid of the snap, crackle and pop.
I understand that will later have to dither down to 16bit from 24bit and down-sample to 44.1.

This makes sense to me on an intuitive level but has anyone ever heard of that practice before and is there any evidence to support that statement or has anyone here even done it and achieved a superior result over just sampling at 44,100?

If it matters, I'll be using Sound Soap 2 and Adobe Audition 2 and recording with an ESI Juli@ sound card unless somebody has a better suggestion for the software.

Thanks!

Increasing Sample Rates On An Audio Recording

Reply #1 –
Quote from: Jimmy The Clam on
I was recently told that if I increase the sample rate (from 44.1k to 88.2 or even 176.4k) of an analog audio recording that I'm digitizing,


I assume you aren't actually changing to analog recording itself but the sampling rate with which you record it?

In that case, yes, if possible, just record at the highest possible sampling rate and bit depth.

Audio editing, specifically very strong editing like removing noise/crackle/whatever, is a lossy process at each step, and when many steps are applied, all the tiny losses do add up. You can minimize the loss by starting from the greatest possible resolution, doing the work, and as a very final step, downsample to 44.1kHz at 16 bit.

Increasing Sample Rates On An Audio Recording

Reply #2 –
Quote from: Garf on
Quote from: Jimmy The Clam on
I was recently told that if I increase the sample rate (from 44.1k to 88.2 or even 176.4k) of an analog audio recording that I'm digitizing,


I assume you aren't actually changing to analog recording itself but the sampling rate with which you record it?


Yes, that is what I meant. Excuse my poorly worded sentence.

Is this a practice that is relatively new in home audio recording/digitizing?
I was just surprised that I had not heard of it before or why more people did not do it.

That is what made me think that maybe it wasn't quite true. Thanks for clearing that up for me.

Do you think that out of either of the two programs I mentioned in my first post that is currently SOTA for someone cleaning up a recording on their home PC, or is their anything else that does a better job while damaging the music less (and does not cost the equivalent of the GDP of Bolivia to buy)?

Thank you

Increasing Sample Rates On An Audio Recording

Reply #3 –
It is nothing new in recording, but new in home recording.

Generally, it is always a good idea to *record* at a high samplerate/bitdepth..... do the editing....... and then as a final step downsample to 16bit/44khz. This has always been the case in digital recording - or even recording in general. Every change to the audio is a lossy process. Thus, you can only decrease soundquality(accuracy), but never increase it. Because of this, it is a good idea to start with a very good source to minimize the editing-losses - and then as a last step downsample.

The only reason why this may not be standard practice yet in home recording is because consumer soundcards which can record at higher bitdepth and samplerate have only become popular since 2-4 years. So, its still relatively new in the home-sector.

- Lyx
I am arrogant and I can afford it because I deliver.

Increasing Sample Rates On An Audio Recording

Reply #4 –
Being on the topic, if a program like WaveLab can record audio from my soundcard at high time and bitdepth resolution, does this mean the sound card really digitises at those settings? I'm just concerned it may simply be the driver upsampling what's recorded by my soundcard, cause it's an onboard cheapo one.

Increasing Sample Rates On An Audio Recording

Reply #5 –
Recording at 24 bit, generally into a 32 bit format, is standard, and I certainly won't recommend against it. Chances are extremely small that you will ever notice an iota of difference when the source is LP or cassette, but it's a reasonable enough practice.

There is a theoretical benefit of recording at a higher sample rate. Do a search on alias image if you want to read about the testing I've done. These effects are real but very probably makes no practical difference. You will never hear, and almost certainly never measure, anything different from an LP or cassette because of it. These aliasing considerations are the only possible reason for using a higher sample rate if your target is CD.

The most reasonable first step after recording is downsampling to 44.1kHz before any thing else. No, working at a higher sample rate will not make the noise reduction, or click and pop removal, or anything else you may decide to do, work any better, but experiment and convince yourself.

No matter what you might capture at a higher sample rate, or what you might do by processing with it, you will have zero effect or benefit from it once you do downsample for CD, so do the downsampling first and save yourself a lot of processing time and disk space. If your target is some higher sample rate media, then perhaps record and stay at that sample rate rather than upsample at the end of processing, although you won't hear any difference either way you do it.

If the soundcard does not have the hardware support for  greater bit depth or sample rate recording, then upsampling will have to occur before the data reaches the program. CreativeLabs cards are infamous for this; many do it even though Creative has often claimed otherwise.

If your soundcard is one that resamples everything to 48kHz on the way through, it is probably beneficial to record at 48kHz and resample only in software. It may be hard to tell with most music, but using the proper test tones really gives one a graphic view (with the proper analysis softwre) of how messy the hardware resampling really is.

Increasing Sample Rates On An Audio Recording

Reply #6 –
Agreed. However I've read that increased sample rates make declicking algorithms work better.

As to bitdepth, in case of LP the noisefloor is so high that recording at 24 bit or 16 bit doesn't matter. I say it doesn´'t matter because 16 bit won't lose anything, and because for post-processing you should work with a 32-bit floating point file, that doesn't matter if was obtained from a 16 or 24 bit recording process.
 

Increasing Sample Rates On An Audio Recording

Reply #7 –
I agree with much of what has been said in this thread. Since the original poster is asking about recording LPs and doing some declicking/decrackling, there are a few additional issues to be pointed out:

1. Some have people have explained that every processing step will involve some losses. However, declicking and decrackling are not "global" effects (like EQ). They operate on discrete sections of waveform. The amendments they make to the waveform at those points are very large - far more significant than any degradation from global effects. In other words, the effect of removing a click is overwhelmingly greater than the quantisation effects of operating at a shorter word length. Thus, recording at greater bit depths is irrelevant in the context of declicking.

2. It has been stated that declicking algorithms work better at higher sampling rates. This is debatable. If a click is audible, then it will have some frequency component below 20kHz, and to a certain extent it is actually better to exclude the higher (inaudible) frequencies from the data in order to minimise the amount of additional data that might complicate the declicker's task. If the click contains only frequencies above 20kHz, then it's not going to be audible anyway.

3. If you do decide to record at higher sampling rates, and the plan is to eventually downsample to 44.1kHz, then it would be better to record at 88.2kHz rather than 96kHz, since this considerably simplifies the downsampling operation. (This would not hold for soundcards which resample to 48kHz internally - as AndyH said, record these at 48kHz and downsample in software).

4. Finally, you will find that there are a number of worthwhile restoration tools which only work at 16 bit and 44.1kHz (and sometimes 48kHz). If you record at higher rates or word depths, you will be eliminating the opportunity to use those tools.

My advice is to stick to 44.1/16 for recordings of LPs.