Vinyl recordings, correct level, amplify/normalize, sample r

Special Interest Groups Recording Techniques
1

Hi everyone

For a few years I’ve been making recordings of my vinyl records using a hifi CD recorder (Sony RCD-W100) connected through a DJ mixer (Pioneer DJM-600). But I decided to upgrade my DJ setup and “go digital” with a Native Instruments Traktor Audio 6 soundcard, which meant I could sell my mixer and cd recorder to clear some space.
Also making recordings without a mixer in the way should mean a theoretical improvement in accuracy as well as meaning I don’t have to rip a CD each time.

Audacity version is 1.3.13-beta (Unicode)
Computer is Windows Vista 32bit
Turntable is a Technics SL1210mk5 with an Audio Technica AT120E cart, aligned with a Baerwald arc protractor
Sound card is a Native Instruments Traktor Audio 6… http://www.native-instruments.com/#/en/products/dj/traktor-audio-6/?page=1964
It’s designed for digital DJing but the specs look good and it has switchable phono inputs. The inputs are switched to phono mode.

I’ve been testing this out for a few days and I’ve got some questions.
I just want to make sure I get my workflow nailed before I start working through my records.

Here’s a screen shot of Audacity whilst recording… http://www.highbeeches.plus.com/recaudacitylevel.gif

  1. Is my recording level acceptable?
    Is that to be expected? Or does this point to some sort of hardware problem or some hidden volume configuration in Windows Vista that I’ve not found yet?
    I’ve got all level sliders that I can find set to the maximum but the peaks of my recording still seem to be <0.5
    Here’s a screen of the Vista Recording Devices property page showing the level set to max, this slider is linked to the one in Audacity… http://www.highbeeches.plus.com/recinchannelproperties.gif
    Previously I used to increase the gain on my mixer so that the peak was just below clipping, then set the recording level on my cd recorder to the same. But it seems with this sound card I’m already at the maximum.

  2. I guess I can artificially raise the level using either the amplify or normalize effects per this wiki… http://wiki.audacityteam.org/index.php?title=Amplify_and_Normalize … since it appears I’ve got it set to the maximum possible level at every stage I can find.
    But by doing that am I losing quality, or making my recording sound worse?
    Because by increasing the level, I’ll also be increasing the level of any distortion relative to silence.
    Or is applying those effects (even to the extent that I need to) just an accepted part of ripping and archiving vinyl records?

  3. I’m not an audiophile mentalist but I like to get the best result I can with the gear I’ve got. The NI Audio 6 sound card supports a 96kHz sample rate and 16bit/24bit format.
    Should I bother recording at a higher sample rate/format?

  4. What’s the best lossless format to export?
    I normally keep all CD rips in FLAC, can FLAC support 96kHz? From the export options it seems FLAC can handle 24bit.
    I will also keep a 44.1kHz/16bit version for more convenient playback.
    But for archival purposes I think it makes sense to go high-res since I’m bothering to do this and my hardware supports it.

  5. I’ve got the sample rate set to 96kHz in the Traktor sound card’s proprietory control panel utility, there’s no option for bit rate.
    But do I also need to set this in Vista’s Recording Devices Advanced Properties for my input channel per the options in this screenshot… http://www.highbeeches.plus.com/recinchannelpropertiesadvanced.gif

  6. Audacity supports 32bit float sample format but my sound card only supports 24bit. So I guess going with 32bit float is unecessary as the information coming in is 24bit, is that right?

  7. When changing the “Default Format” under “Advanced Settings” for my input under “Recording Devices” in Vista to:
    2 channel, 24 bit, 96000Hz (Studio quality)
    The recording level is almost zero, slight 1-3 pixel peaks are seen on the graph.
    But if I change it to…
    2 channel, 16 bit, 96000Hz (Studio quality)
    Then it comes through fine.
    Does that mean I need to also tell the soundcard hardware to output 24bit instead of 16bit?
    Or does it mean that there’s some other problem?

  8. Is there a way to set the output/input level monitors to default to linear instead of dB scale? I have my tracks set to default to linear in Audacity’s prefs.

  9. Software playthrough. I don’t have a hardware playthrough option so software playthrough is the only way I’ve found to be able to hear what’s being recorded.
    My machine seems to be able to handle the load, it’s a chuddy Toshiba laptop with 4GB memory and an Intel dual core T2250 at 1.73GHz.
    But is it recommended to turn that off to prioritise the recording?

I realise these questions must get asked all the time on here. I’ve only ever used Audacity for chopping around existing tracks. My recording use of Audacity was with an earlier version and under XP so things have changed quite a bit since then. It seems the sound device options for Vista are spread all over the shop but something that appears to be much improved in Win 7.

And I guess everyone’s setup is always slightly different, with different combinations of computer hardware. I just want to make sure I’m not missing anything so I don’t find out in 18 months’ time that I missed a checkbox somewhere that prevents me getting the best output I can within my time and budget.
I’m also not bothered about removing any pops/clicks from my vinyl recording. To me that’s part of the recording and could always be done at a later stage.

Thanks in advance!
Cheers, B

2

Nice to see so much detail, it really helps :slight_smile:

It looks a little low, but acceptable.
From the web site info there does not appear to be any gain control on the Traktor Audio 6, and the quoted 29.1 dB gain is a little on the low side, though some DJ 12" singles can be very loud and, as you say, the unit has been designed with DJs in mind.

That’s possibly a little on the high side when recording in Audacity. A CD recorder may include circuitry to limit the maximum input level, so reducing damage if a few peaks go over 0 dB. In digital audio 0 dB is an unforgiving limit. CD audio can not go over 0 dB, so if an analogue signal tries to go over 0 dB it will clip the top/bottom of the waveform, which is a nasty kind of distortion. When recording on a computer, it is a golden rule to keep the peak levels below 0 dB. As long as you test the recording level with the loudest part of a recording you can probably push the recording level up to -3 dB safely, but much higher than that there’s a risk of getting a few peaks that mess up the recording. (for live music recording it is usually necessary to leave a lot more headroom due to the unpredictable nature of live music).

The problem with recording too low is that there are less “bits” to describe the sound in digital format so the amount of noise relative to the maximum music level may increase. In practice you have to go quite a long way down before it becomes a problem. Even if the peak level is only -18 dB, the digital format still has at least 78 dB range. From your picture it looks like your peak level is around -8 dB which should be fine.

There are pros and cons.
Unfortunately on Windows Audacity cannot capture audio above 16 bit (yet) so there are no obvious benefits to switching to 24-bit. (There ARE benefits to setting Audacity to record in 32-bit float as it improves the quality of any processing).
Increasing the sample rate has the theoretical benefit of increasing the audio bandwidth. On the other hand, there is double the amount of data which makes everything work harder (disk drive, processor, A/D converters …).

There may also be subtle differences due to the sound card design.
I’d recommend that you make a few test recordings. If higher settings sound better, then use them. If there is no audible difference, or lower settings sound better, then use lower settings. CD audio is 16-bit 44100 Hz, so you don’t want to go lower than that. Recording at 44100 Hz has a benefit that if the final version is going to be on CD, then the sample rate will not need to be converted. The sample rate converter in Audacity is excellent, but there is always a tiny loss in quality (inaudible, but still there).

The very best format is "32-bit float PCM WAV at whatever sample rate you are using in Audacity. These files are huge.
The next best is “24-bit” (uncompressed) at the same sample rate as the project. Here FLAC is the clear winner due to the smaller file size.

Probably yes. Whatever sample rate you are using, set it the same everywhere you can.

If you are doing any processing (Amplify, Equalization, Click Removal, Fade Out, ,…) then 32-bit float is better and usually faster.

No one accuses Microsoft of making bug free sound card drivers. Don’t use 24 bit, 96000Hz.

Not that I know of.

If there’s a problem you will hear clicks and/or skips in the recording when you play it back. If the recording sounds good when you play it back then you can safely leave software playthrough on. Note that software playthrough is delayed a little (typically around half a second) before you hear it. This has no effect on normal playing back what you have recorded - it only applies to what you can hear during recording.

There’s a lot of relevant information in the manual which you can find here: http://manual.audacityteam.org/man/Tutorial_-_Copying_tapes,_LPs_or_minidiscs_to_CD

3

As Steve says …

This workflow tutorial from the manual may help too: http://manual.audacityteam.org/man/Sample_workflow_for_LP_digitization

WC

4

Hi guys
Thanks for the replies. Apologies for not getting back on this sooner.

Yeah that’s what I figured. Recording low means the ratio of noise to sound is much closer and that difference would remain constant if I increased the level.
So should I amplify or normalize the recording and to what level?

Been doing a bunch of “final” recordings at CD quality, which was fine. Although some of the records suffered from quite bad sibilance though I think that’s either the record or most likely my alignment of cartridge in the tonearm. I’ll be getting some further opinions and advice on this from a friend in a week or so.

I am also completely fine with using up extra disk space and resources to archive 96kHz 16bit versions and keep a 44.1kHz 16bit for convenient playback.
The fact that Audacity doesn’t support audio capture above 16 bit could be the reason I had problems when setting Vista to 24bit and it only capturing sound at a tiny volume level.
Also there is nowhere in the driver software for me to be able to select between 24bit or 16bit. So I assume that the internal electronics of the sound card operate at 24bit then downsample to 16bit once it arrives at the computer.

I think my best bet is to go for 96kHz at 32bit integer, then save as 96kHz 16bit WAV.
Can I convert that 96kHz WAV to a 96kHz FLAC? Or does FLAC not support 96kHz?
Then I’ll convert that to a CD quality WAV for more convenient playback. Though a temp MP3 might be easier as I always have the high-res to re-export if I need to.

Cheers, B

5

Yes you can.