I record using ProTools and have just completed an extremely lengthy, time-consuming session - only to discover the sample rate was set at 44.1Khz! Does anyone know how important this really is? There is a chance this track will be commercially released and I understand that 48Khz is the industry standard. I just want to know if I'm going to have to re-record the track or whether that will just be a waste of time!

Er...what format is the release?

CD or DVD?

yeah *EVERYONE* who hears the results will INSTANTLY recognise it was at 44.1kHz! It will be SOOOOOOOOOOOOOo obvious.

(not)

--------------------
~~~ A weasel hath not such a deal of spleen as you are tossed with! www.feline1.co.uk ~~~

no problem....lots of people work at 44.1k as well as 48k.

J

--------------------
www.jackruston.com

That's good to know!

Re-recording the whole thing would have been a nightmare - I don't think I can tell the difference - just wanted to be sure!

Quote Yonatan:

---

I record using ProTools and have just completed an extremely lengthy, time-consuming session - only to discover the sample rate was set at 44.1Khz! Does anyone know how important this really is?

---

In the grand schemes of things, it isn't very important. You can easily sample rate convert to 48kHz without losing anything, if really necessary.

Quote:

---

There is a chance this track will be commercially released and I understand that 48Khz is the industry standard.

---

It is for anything related to video/film pictures. But CDs are at 44.1kHz.

Quote:

---

I just want to know if I'm going to have to re-record the track or whether that will just be a waste of time!

---

It would be a massive waste of time. Use a sample rate converter if you need to change the rate up to 48kHz.

Hugh

--------------------
Technical Editor, Sound On Sound

yeah it would definitely be a waste of time - especially as (even still) the majority of sessions are done at 44.1 !!

--------------------
Battenburg to the power of 20 - said by Richie Royale in a moment of genius. 4pm. Wed 16th Nov 2011. Remember where you were....

If you are working for cd you're fine but it's better to work on 88.2 sampling rates and than resample it on 44.1KHz.If you're doing sound for image than you should resamle your work to 48KHz.

--------------------
nikolavm27

There's an easy way round this. The difference between 44.1khz and 48khz is only 3.9khz. So if you record 3.9khz worth of silence and add it to the end of each of the audio files recorded at 44.1khz, then that will make them up to 48khz.

Sample rate converting to 48khz might work as long as the audio wasn't recorded at 24 bit. 48 = exactly 2 x 24 so in that case you would create terrible problems with room modes. It would also cause one of your speakers to resonate at twice the frequency of the other which would badly distort the stereo image.

Sorry am I the only one here who thought this thread was a pisstake?

Quote Wurlitzer:

---

There's an easy way round this. The difference between 44.1khz and 48khz is only 3.9khz. So if you record 3.9khz worth of silence and add it to the end of each of the audio files recorded at 44.1khz, then that will make them up to 48khz.

Sample rate converting to 48khz might work as long as the audio wasn't recorded at 24 bit. 48 = exactly 2 x 24 so in that case you would create terrible problems with room modes. It would also cause one of your speakers to resonate at twice the frequency of the other which would badly distort the stereo image.

---

I'm pretty sure Russ Andrews makes a box that will do this.....

--------------------
Cubase, guitars.

Quote nikolavm27:

---

If you are working for cd you're fine but it's better to work on 88.2 sampling rates and than resample it on 44.1KHz.If you're doing sound for image than you should resamle your work to 48KHz.

---

Hi nikolavm, welcome to the forum. That informaion is very much out of date. Modern sample rate conversion does not require the sampling rates to be simple multiples of each other. The program will just multiply the rate up to some huge number until it gets a multiple, and then it converts.

--------------------
madtheory creations
Synths and pianos for Kontakt

Quote Wurlitzer:

---

Sorry am I the only one here who thought this thread was a pisstake?

---

Well done Wurly. You made me laugh quite a bit. I had to check who posted your comment. I thought it was serious at first!

SJ.

newbie, so i'm really only guessing here, BUT.... if recorded at 44.1 then converting the sample rate doesnt make any difference to the sound AT ALL unless you are down-sampling. All the conversion does is show the file to be 48 instead of 44. ????

--------------------
That wasn't a wrong note, just a bad choice!

Quote Wurlitzer:

---

Sorry am I the only one here who thought this thread was a pisstake?

---

well.... no...

No, converting does actually do something. I believe modern sample rate convertors are very transparent, that's what is meant by "won't make any difference to the sound".

yeah a good sample rate converter is supposed not to make any difference to the sound....

--------------------
http://www.myspace.com/joystickmusicgreece

Sampel rate conversion is exactly that. A conversion of the sampling rate. From 44.1 to 48 or whatever.

if you are increasing the sample rate, there is no change to the equivalent (analogue) audio information as such, but the data describingh it is changed to corerspond with the new sample rate.

If you are downsampling, then the audio information has to be changed to ensure it complies with the Nyquist rules. For example, a recording made at 96kHz can, in theory, hold information about audio frequencies up to about 48kHz. However a 44.1 kHz sampled signal can only hold information on frequencies up to about 22kHz. Consequently, part of the sample rate conversion process must also include a low-pass filtering stage to remove everything above 22kHz.

Ten or more years ago, sample rate converters used to work much as Tomas suggested: take the source signal, oversample it to a ludicrously high rate, then down convert. Ideally, the sample rate of the oversampled intermediate would be a common multiple of both the source and destination rates.

However, often there is no common multiple that is both convenient and practical, and this appropach can't work at all if the input rate is varying (such as with a varispeed source). So instead, most early systems simple picked a constant large oversample factor (typically 256x). In the case of a 44.1kHz source this gives intermediate oversampled stage running at around 12MHz. What they then did was pull out the samples closest to those required by the destination rate clock. Some might be perfectly in time, most wouldn't be, but the timing errors are small...

But what we have here, inherently, is flawed system. Samples are being pulled out of the intermediate stage that aren't quite inthe right place, timewise -- so we have a form of jitter. That results in quantising errors and noise. That's why these early converters were mostly limited to 16 bit performance. The internal errors made it impossible to go beyond that kind of resolution.

These days a very different approach is used. The oversampling idea described above is very processor-intensive but not very clever. The system is doing a huge amount of work to generate the very high oversampling rate, but most of the calulated samples are discarded. You start with 44100 sample a second, calculate 11289600 samples a second, and then discard 11241600 of them to leave the 48000 you wanted.

Modern SRCs are still very processor intensive, but they apply the processing in a far more intelligent way. Essentially, they compare the relative phase and rates of the incoming data and the output clock, and use that information to work out when an output sample will be required, and then calulate the value necessary for that specific sample. As a result, the accuracy is considerably higher, there is no jitter effect or quantisation error. Most modern SRCs are accurate to well beyond 24 bits and their performance is well beyond that of even the best D-A converters -- hence my claim that they are transparent in what they do. Assuming the filtering and everything else is performed correctly, you won't hear any degradation from a pass through an SRC.

Most high end digital consoles these days have SRCs on their inputs all the time, for convenience, without fears of reduced signal quality.

Hugh

--------------------
Technical Editor, Sound On Sound

Thanks Hugh - very informative.

When I started digital recording, I was advised to record at 44.1 because the SRC thing wasn't too great back then. Things seem to have moved on a bit, so is it now worth recording at 48 because to take better advantage of A/D converters - I understand there are filtering issues because at some point a converter has to roll off the high end to prevent alaising. At 44 it can start in the audible hearing range. 48 hopefully starts the roll off outside of the audible range. That's the theory at least.

Does this mean I should start recording future projects at 48 rather than 44.1, and then SRC down (or get the mastering engineer to do that?).
Any programs in particular recommended for SRC (quality seems to vary)? And also for dithering (24bit to 16bit)?

Ta,
D