Recording using digital input

Hi

I am new to Audacity but have been impressed with what I have seen so far.

I am in the process of transferring my audio collection (CDs, cassettes, LPs) to a hard drive on my PC. I have used foobar 2000 to rip the CDs and save in FLAC format and am just about to start copying cassettes, hence downloading Audacity. There is one particular aspect that I don’t currently understand and would welcome an explanation of, but before I come to that I’d better explain the setup I’m using.

My computer has the cheapest soundcard I could find that came with digital input (Trust SC-5250 5.1 Surround Sound Card). The analogue output from my cassette deck is connected to my Minidisc player (NB I am not recording to minidisc, merely using the player as an analogue-to-digital converter), then using a optical cable to connect the digital output from the Minidisc to the soundcard input. (My assumption is that the Minidisc analogue-to-digital converter will perform better than that in the soundcard - any comments?) I am using Audacity 1.3.4-beta.

When I record from cassette into Audacity (using the setup described above) the track is described as Stereo, 44100Hz 32-bit float. My concern is that if this involved a change of sample rate then there will have been data loss - I want the recordings on the hard drive to be the best quality possible, even if my current method of playing involves some degradation. So, how can I find the format of the output from my Minidisc player and thus ensure there is no sample rate conversion?

I also have a comment about the speed of the FLAC compression as performed by Audacity. When using foobar, the ripping of a 60 minute CD, including saving as FLAC files, seemed to take about 10 minutes (I haven’t performed proper timing). However, on a short trial of maybe 5 minutes of cassette recording using Audacity, the initial input to Audacity obviously took 5 minutes (I can only play the cassette at the proper rate) but the saving to FLAC seemed to take a further 5 minutes. Why should the Audacity FLAC compression take ‘real-time’ whereas that in foobar must be considerably faster?

As a further consideration for the future, I have previously copied my LPs to Minidisc, thinking that would provide long-term storage and convenient medium for playback. I have now decided that the future lies with saving the audio on hard drives. So, I have a choice when I come to copying my LPs. I can either go back to the original LPs and use a similar setup to that for cassette, or I can copy from the minidiscs, which would be a lot more convenient. My question here is whether I’d lose much quality by copying from the minidisc rather than from the LP.

I look forward to your response.

Regards
Dave

The A/D converter on my minidisc is very good and considerably better than on a budget sound card.

Minidisc uses 44.1 kHz sample frequency.
S/PDIF uses 20 bit data (occasionally 24) but it is possible that the minidisk may output it as 16 bit and add “padding” (zeros) for the extra 4 bits - You will need to see the minidisk documentation to be sure (if it exists).

I’ve just run a test on my old Pentium 500:
10 minute stereo 44.1 kHz/16 bit to 16 bit FLAC
Medium setting (4) - 21 seconds
Highest setting - 1 minute 31 seconds

When you Export a track in Audacity, multiple data files have to be accessed and pieced together to create the mixdown. Depending on how much editing and processing you have done, this could possibly account for the extra time that you are experiencing.

Minidiscs use “lossy” compression, but if the sound quality when you play your Minidiscs is good enough for listening to, then there is likely to be little difference between the minidisc and a CD made from the Minidisc.
If your LP’s have been in constant use, the old minidic recordings (of brand new records?) may now be better than the records.
Best thing to do here is try it and see (hear).

Thanks Steve, that is just the sort of response I was hoping for. I now feel more confident about going ahead with my recordings.

With regard to the speed of export, I certainly hadn’t intentionally performed any editing or processing but I will experiment a bit more , and maybe even read the instructions, to make sure.

Even though I am now confident that I ought to be using 44100Hz as the sample rate my curiosity is aroused about the circumstances under which Audacity performs a sample rate conversion. I would have expected input to be stored at the sample rate/bit depth that it arrived at, conversion being performed only when using playback or export. I need to understand better (and will look at the User Manual) what governs way the input is initially stored and the effect changing the project and/or track rates has on the stored data.

Thanks again.

Regards
Dave

I think that Audacity will record at the project rate, but if you import audio, then the sample rate is kept the same and only converted for playing (not changing the actual imported data). When you Export or mix down, sample rate conversion is applied as necessary according to the project sample rate.

Perhaps when you have looked into this, you could post your findings. (or if anyone else knows?)

I’ve now done some research on sample rates.

Firstly, I did some tests using an analogue input into my sound card recording approximately 30 seconds of stereo audio. With the project rate set at 8000Hz Audacity used about 2MB of disk space to store the files (NB I am not talking about exported files, these are Audacity’s temporary working files). With the project rate set at 96000Hz the disk space was more like 18MB (I guess it should actually have been 12 times the amount, but something in my computer was failing to keep up with the data rate - as I don’t intend to use this rate in reality then it doesn’t really matter whether it is the sound card, the processor or the hard disk that is causing the problem). My supposition at this point was that when the project rate was at 8000Hz, say, then Audacity instructed the sound card to convert analogue to digital at that rate, and then Audacity recorded what it received, without performing a rate conversion. Similarly, for the 96000Hz rate, the sound card was instructed to convert at that rate and, again, Audacity was not performing a conversion on the data it received.

Next I did a couple of tests with digital input from a Minidisc player, so the input to the sound card is digital at 44100Hz (thanks stevethefiddle). Again I recorded about 30 seconds of stereo audio with the project rate set at 8000Hz. As in the analogue case the file size was approximately 2MB. Then I did the same with the project rate at 96000Hz, the file size was then about 18MB (again something in the computer was failing to keep up with the data rate).

The results with digital input surprised me. I had made the following assumptions:

  1. Audacity does not perform rate conversion on recording (this is implied by the ‘Set Rate’ section on http://audacityteam.org/manual-1.2/track_popdown.html)
  2. Audacity instructs the sound card to convert analogue to digital at the project rate when beginning recording. (Is there any way to tell what rate the sound card is outputting?)
  3. The sound card would not perform a rate conversion on digital input but would just pass it straight through.

However, if these assumptions were true then the files recorded from digital input should be the same size regardless of whether the project rate was 8000Hz or 96000Hz.

Comments welcome!

PS I have only just noticed that, during and after recording, at the bottom right of the Audacity window there is a notification of ‘Actual Rate’. In my case this always appears to be 44100Hz, for both analogue and digital input. So it now appears that:

  1. My soundcard always outputs at 44100Hz (although I think that it is supposed to be able to output at different rates).
  2. Audacity does perform a rate conversion on recording if the Project rate does not match the actual rate.

A lot of sound cards have a “native” sample rate - that is, a specific fixed sample rate for the hardware.

For (most/all?) SoundBlaster Live cards the hardware runs at 48 kHz, however the standard Windows drivers allow the data to be passed to applications at various other rates. So the hardware + driver, as a system, is able to convert the sample rate. This is not the case when using some other drivers, for example ASIO, which is tied to the actual hardware sampling frequency.

For your sound card, it looks like the hardware uses 44.1 kHz.

The sample rate indicated in the track itself shows how Audacity will treat the data. The recorded track is just a load of numbers, each representing a “sample”. If the track is set at 44.1 kHz, then Audacity will allocate 1/44100 of a second along the time line for each sample. If you change the rate to 22050 Hz, then Audacity will “space them out” more, so that each sample is allocated 1/22050 of a second, so the duration will double and the the pitch will go down by an octave. However, if your “Project sample rate” is different, for example 48 kHz, then Audacity will convert all tracks to the project rate for playback, mixing and export. Changing the project rate and recording will produce data files that reflect the sample rate - for example, setting the project rate to 96000 Hz and recording, will produce files that are twice as big as if the same recording was made with the project rate set to 48000 Hz.

Higher sample rates give greater bandwidth, that is, higher frequencies can be recorded. However, if the audio hardware is fixed at 44.1 kHz, then no matter how high you set the project rate, the frequency range that you can record is still limited by the capabilities of the audio hardware - in this case, about 17 kHz. So if you are doing simple recordings, there is little point using sample rates any higher than you hardware actually runs at. Higher sample rates will just create bigger files and increased demands on you processor and hard disk, with no improvement to sound quality.