Moderators: If this post is in the wrong place, please move it.
I am starting a new thread on this subject even though it overlaps with the thread ‘Ripping Vinyl’ in the ‘Recording Techniques’ section of ‘All Things Audio’. The last append to that thread was on April 19 so I assume that nobody is looking at it any more.
The first topic I would like to ask about (possible again) is recording volume. In the earlier thread “Ripping Vinyl” member RS_RS mentioned using an M-Audio card in his set-up. I have the Delta AP (i.e. 2496 Audiophile) model of that card. After puzzling over it for some months, I finally learned (from an m-audio knowledge base item) that there is no way of controlling recording volume with this audio card.
This becomes a problem when recording from my FM Tuner – input frequently exceeds the guideline of 0.5 on the waveform display in Audacity, or is greater than -6 dB on the VU meters. In contrast, the input is often down at -24 dB when recording from Vinyl. (BTW I’m using a modest Rega MM cartridge – but I don’t think this is the problem, because there are some recordings, or some passages in some recordings, which go to 0 dB).
I am feeding the analogue input on the m-audio card from the only line-output on my Arcam Alpha 9 integrated amp. This is the ‘Tape Record’ output. It is upstream of the Tone and Volume controls, as one would prefer. The only other option is to use the main outputs from the speaker driver stage- and I’m not going to attach that to the analogue inputs on my m-audio card!
So, the only option I seem to have is to control volume after the fact - by use of ‘normalize’ in Audacity to boost or reduce the peaks. But I see 2 problems here: I assume that any one ‘click’ spike will mislead ‘normalize’ and cause it to reduce the volume too much – correct? On the other hand, boosting the average record volume will also boost the underlying hiss and noise. So, should I do all ‘cleaning’ of the audio before normalizing?
The second, simpler question, is: ‘what digital format should I use to hold my digitised vinyl recording, to ensure longevity?’ Actually, there are a number of subsidiary questions, (e.g. about what media to use to store the recording long term), but I’ll wait to see what comments there are.
In my opinion, as long as you don’t get clipping (use View > Show Clipping) you’re fine. Even with a recorded level of -24 dB the analog noise will swamp the digital noise. Yes, you will increase the apparent level of the hiss, but the overall signal-to-noise ratio will not be affected.
If your recording has clicks and pops, remove those before doing Normalize. Either use Audacity’s tools (Repair, Click Removal) or export the recording and process it in a third-party click-removal tool, then import the processed recording for further editing (this is what I do).
As for long-term storage, I am not a fan of using 32-bit as I don’t see the benefit. I use 16-bit FLAC on an external hard disk. I wouldn’t trust CD-R or DVD-R for more than a couple of years. This is a controversial subject and you might want to search the web for information and opinions.
Thanks for the advice - it confirms much O what I was planning to do. I am finding that not many exteneral tools will handle 32-bit float - which is what I prefer to use while ‘cleaning’ the digitsed audio signal, before finally storing it as 16 bit - probably FLAC as you suggest.
I quite concur with your comments on the longevity of optical media. Even though I have switched to using only Taiyo Yuden manufactured media, I still doubt that such recordings will last more than 8 to 12 years (most of my data backups on main-stream media, kept in dark cool environment have so far lasted 5 to 8 years). However, rotating magnetic media have other forms of failure: so far, after 30+ years experience of hard disks, I have not found one which will last for more than 8 years. This does mean that I have to keep 2 copies or more, using some form of RAID to ensure recoverability.
And the super-cautious keep off-site copies too, just in case the house burns down/floods or the cat knocks over the disk - don’t ask me how I know that
Yes indeed, not even attenuation, let alone amplification, as far as the input is concerned. You can use the hardware mixer to attenuate the digitised signal, but that’s not your problem.
To avoid clipping, all that you need to do is to stay below 0dB. The -6dB guideline is simply to allow for the occasional higher-than-expected peak. Are you still in trouble at 0dB? If so, you will need some sort of analogue pre-amp between the tuner and the card, at least with the capability to attenuate the signal with minimal distortion. Presumably in some countries there is still some reason for recording from FM radio. Here in the UK you can get everything from the BBC over the internet, and in the case of Radio 3 at AAC 320kb/s 44.1kHz which is near CD quality. You should look into what is available for you. The problem might go away (but see also the thread on Freecorder).
Good call! I assume your amp doesn’t have pre-amp out connections. If it did, at the cost of a small amount of additional distortion, you could use those to allow the volume control to be applied, with the tone controls defeated or at least centred. But (a) this might not give you anything above standard line-level that you are getting from the tape out connections, and (b) if you are occasionally getting close to 0dB from vinyl, then you don’t want to increase the signal level anyhow.
You can’t recover from digital clipping, so you must find a way of making sure that your analogue input signal does not exceed what the 2496 can handle. As far as having too low a leve of input signal is concerned, what really matters here is the difference between signal and noise. Roughly speaking, if your input level is high enough that noise on the analogue signal is still getting through the digitisation process at a reasonable level, then you are not losing any signal, and normalising is just a matter of proviiding a convenient level for your final digital version. Pre-amplification risks adding distortion with no compensating benefit. If you want more signal, you may need a change of cartridge and/or a separate phono pre-amp like the NAD PP2.
Your 2496 will digitise at 24-bit nominal depth (which will in practice probably give just a bit more than 16 bits of data with garbage in the least significant bits) but Windows will in any case truncate this to 16 bits via either MME or Direct Sound… You can get at the full 24 bits by compiling an ASIO-enabled version of Audacity (see my recent thread in the “Compiling Audacity” forum) and using the ASIO drivers for the 2496. Works for me, just gives that bit more headroom for post-processing.
Like a number of other Audacity users, I export my raw transcribed files as 24-bit .WAV files and then use DeNoiseLF, ClickRepair, and DeNoise to clean them up, before re-importing into Audacity to adjust volume, split onto tracks, etc. You definitely do not want to use Normalise or Amplify while you still have high-amplitude nasties in the data.
As others have said, FLAC or, if space is not an issue, uncompressed WAV shuld be a very safe bet. A big disk on your machine and a portable disk kept somewhere else (ideally, in a disused mine, but that may be over-engineering the arrangements) is the usual recommendation.
