need advice on ideal recording settings

Greetings, I recently purchased my first audio interface; it’s a Phonic Firefly 302 USB. I work for a small and relatively new archive, making digital copies of old records and tapes. Here is my question: What are the ideal settings in Audacity to achieve high quality, archive-grade recordings of old records, reel to reel tapes, etc. Here’s what I’ve currently got: The Phonic Firefly processes at 24-bit, and I have its sampling rate set at 96kHz. In Audacity, I’ve set my project rate at 96kHz and 32-bit float. The real-time conversion settings are “Fast Sinc Interpolation” with no dither, and the high-quality conversion settings are “High-qulaity sinc Interpolation” with shaped dither. I import the recordings as AIFF files (instead of WAVs as AIFF files allow the incorporation of metadata into the file itself) at “Signed 24 bit PCM”. Is this the best combination of settings, or do I need to fine tune them? I’m relatively new at this, so any input would be much appreciated. File size is not an issue; we’re just going for the best digitized recordings we can get.

On a separte note, I thus far really like the Phonic Firefly. My only concern is that even with its input knobs turned all the way up, I’m only getting a maximum of -3dB as read by Audacity’s input meters. Quieter recordings max out around -12 dB. Should I be concerned about this? I realize that I can amplify the recordings ex post facto with Audacity’s Amplify effect, but would the end result be better if the signal were louder at the time of its recording / capture?

Thanks in advance for any advice!


Even though modern music productions are dense and seem to live smashed up against “0,” that represents a great deal of theatrical processing in a studio. Live and archival recording doesn’t do that.

This is a nearly perfect live capture recording. Note that most of the music stays below 0.5 which is -6dB.

Before you fall in love with your sound device, what is the noise floor? Set up for a recording, but don’t press play on the player. Where does Audacity settle? We’re expecting “grass” or very low level noise around the -60dB mark. Music lives between that and overload or clipping at 0dB. Even though there are tools to “cure” clipping overload, they don’t really work all that well, so overloading is a very bad thing.

Drag-select some of the noise.
Effect > Amplify > [enter].
Listen to it. It should sound like rain-in-trees shshshshshshshsh. If it’s buzzy or any other harsh sound, then there is interference with the music – generally digital problems.

Does your device get its power from the computer – does it have a wall brick supplying power [pause] Yes, I see it does. Good. The all-in-one USB devices have problems with computer noise.


Thanks for the great advice, Koz. I’ve doing some floor noise checks on a couple of different components like you suggested. On a minidisc player the floor doesn’t even show up on the input meter, so I’m assuming it’s below -60 dB. When amplified, I get a continuous, constant shshsh sound you described. The wave form, when amplified, is short and without variation. On a turntable, the floor bounces between -60dB and -50dB. I still get the contant rain in tree sound when amplified, but the wave form is more variable with peaks and valleys. Any cause for concern?

Personally for my archival of LPs, tapes and Minidiscs (and for current live capture from BBC FM broadcasts) I have my Audacity set at 32-bit float 44.1kHz for recording and editing - the 32-bit float gives good headroom for editing functions. And then when I produce the finished WAVs I use Audacity to downsample to 16-bit PCM stereo WAVs using triangular dithering. These settings produce very good results for me - I have good playback equipment, but my ears are ageing.

Serious archivists may well choose higher settings but I find the above fine for creating good playback quality for CDs and iTunes uploads.

But yes I do understand why you are asking, it took me a lot of research when I started out to decide the parameter settings that would work well for me.


I was intentionally waiting for another elf to suggest data parameters. I have no good preferences for this.

“Noise” is interesting because white noise, by definition is random crap at all frequencies (so to speak). The lumpy one is probably more “natural” than the flat one, but both appear to be perfectly normal.

The phono channel is most likely to hum because the intentional processing in the amplifiers encourages it. If that one’s quiet, you’re probably good to go.

There’s a toss-up between Just Enough Data and Dramatic Overkill. There are advocates for both. Music CD specifications are Stereo, 44100, 16-bit. Some would say more than that is a waste, but that’s barely adequate if you want to apply filters and effects – like clean up the pops and clicks on your vinyl. One goal is to end up at 44100, 16-bit, not start there.

On the other hand, very high specifications will only amplify radio signals from the airport and not contribute to the show at all, plus those files take up an enormous amount of storage.


I’m a little confused by this statement as I thought you were “recording” - or do you mean that you “Export” as AIFF Signed 24 bit PCM?

Sorry, Steve, I indeed mean “export,” not import.

That’s what I thought.

A couple of things to consider:
Regarding the Export format, it really depends on what you want to do with the recordings. To export a file that is a perfect copy of Audacity track(s), export as “32-bit float PCM WAV”. Exporting as 24-bit is probably so close to perfect as to not make any difference (apart from smaller file size). The disadvantages of 24 or 32-bit formats is that they are much less widely supported than 16-bit. Audio CDs are always 16 bit (integer). Provided that the audio is amplified close to 0 dB prior to exporting, there is little difference in audible sound quality between any of these formats though you may notice that for 16 bit exports, “silence” in the recording is not totally silent. For making CDs this is probably not important as CD players will often mute sounds when they drop to the 16 bit noise floor. For archiving purposes, 24 bit Flac is a good option as it provides full (uncompressed) 24-bit quality, but with some saving on file size. It also has good support for metadata. Again, there are not many programs that support this format, but it is quick and easy to convert Flac to PCM.

Other than this final decision on the export file type you look good to go :slight_smile:

Thanks, Steve. I actually catalogue AIFF files in iTunes, which allows us quick, easy, and organized access. I just did a test to see if iTunes would accept & play an AIFF file exported at 24-bit, and it seems to do so without problems. Although I haven’t tried this yet, I’m assuming that iTunes would be able to convert such a file from 24-bit to 16-bit when making an audio CD - is that a correct assumption?

Try it. I know iTunes will cheerfully convert between 48000 and 44100 to manage video sound files, so it should be OK. iTunes uses QuickTime Services and they recognize a broad range of formats.

But try it out.

Also, iTunes can be used to create compressed versions of the music. Set that in CD > Importing (I have no idea why they put it there). Default is AAC High 64. I use AAC Plus 128. You can pick MP3 or whatever is in that list.

Select the song. > Advanced > Create AAC (or whatever) version? It will make a copy of the song with compression.

Thanks to everyone for the helpful assistance. I do have a follow up question regarding Audacity’s Amplify feature. Let’s say I’ve captured/recorded some audio that has a very weak signal, maxing out around -24 dB. Would amplifying the recording (without allowing clipping, of course) in Audacity introduce any noise or distortion that wasn’t in the recording to begin with? I fully realize that it would amplify the noise already present in the original source; I’m just wondering if it would in any way degrade the quality of the finished product. This is assuming that Audacity is set at 32 bit float 96kHz for the recording and amplification.

The sample rate is not relevant to this question, but the bit format is.
For audio tracks that are in 32-bit float format you can amplify almost as much and often as you like.

Of course if you take it to ridiculous extremes of amplifying by many hundred dB then it is possible to cause damage, but for a more normal range (say within +/- 50 dB) you can amplify the level up and down hundreds of times without causing damage. To put it another way, if you amplify a 32-bit float track so that it is 1 million times quieter, and then amplify it a million times louder so that it is back up to it’s previous level, then 32-bit float can handle that with no noticeable noise or distortion.

Ok another recommended parameter …

As Steve says, with 32-bit you can amplify to your heart’s content - however I would not recommend Amplifing all the way up to the maximum 0dB - I would recommend choking it back to -3.0dB as your max level with Amplify - this is because some players struugle with a full 0dB signal and any -3dB is plenty loud enough.

When I am recording from LP I aim for a mazx signal level of -9 to -6 dB, do all my editing and than as the final step I “normalize” my recordings up to -3.0 dB prior to exporting the WAVs.

BTW don’t be tempted to use the actual “Normalize” effect for this - Normalize in Audacity currently operates on each stereo channel independently and thus can damage your stereo balance and your stereo image (don’t ask !! ). You may find this article in the Wiki interesting:

_There is a proposal being developed to change that behaviour, but don’t expect to see it implemented any time soon. See: and