Like many young ambitious DJs on the digital side, I scrambled early on to gather the largest library I could hold. As time moved on, I started to pay more attention to audio quality and more recently I’ve been nearly obsessive about it.
Lately I’ve started digging through my old library and playlists, finding little gems that I’d love to toss into a mix here and there, but I’m afraid to muddy a great mix with a bad-quality track. I know I can take the long ‘cheat’ route of comparing file-size to length of track, but that feels cumbersome and a little imprecise.
Is there a software available (or a certain method) that anyone would recommend to validate audio quality without having to listen, track by track, trying to make judgement calls?
Stop it! Stop it right now! Do not give into the dark side of audiophile-issnicity. Back away from the edge right now!
You can display the bitrate of mp3’s in most software an a column of information that you can select. This can be setup in the file broswer, iTunes, (almost) every DJ program, etc. The details will vary with the software and OS…but the information is there. Google is the best place to start. “…how to display the mp3 bitrate in ____…”
I have a HUGE collection of music going back to the mid-90’s. In 1996-ish I ripped mp3’s to 160kbps. That was a reasonable balance between file size and quality given the cost of disc space and the specs of the audio interfaces (hard to imagine today, but disc space was $$$$ back in the day). All my mp3’s that are below 192kbps are tagged with “LowBitrate” in the comments. I can exclude those when making smart playlists in iTunes…but I still have them in a pinch.
I used to dl Mp3’s religiously until I played on a 50K system. Wav or flac only from now on. The lossless are much more full, mp3’s creat artifacts and actually get rid of certain frequencies.
For analyzing songs I use this program.
DL and then just drag any song onto the program and it analyzes the freq range and file size. 320 Mp3s are cut around 20khz and heavily degraded music around 13-15k hz.
Lossless is better, and given the price of bandwidth and storage, should be the “preferred” solution for DJs. For those of us with extensive collections of mp3’s (and no good way to purchase lossless replacements), easily identifying the bitrate of the mp3 is a useful thing to do.
So is my hearing after 20 years around PA systems. Perspective helps when making these kinds of decisions.
Just buy the tracks you love and get rid of everything below 320. Ask yourself if you love those tracks enough to buy new HQ version of it. If you dont, then you dont actually love them and can throw them out.
Your library will sort itself out, and be a lot smaller and a more accurate representation of your taste.
dBpoweramp is a bit better to use in this regard. It gives you a context menu for audio properties, which will tell you not only the bit rate, but also the codec used to rip it. What you really want to see here is that LAME was used to encode the MP3. You’ll still have to open the folder and right click the file, which takes a bit of time.
I believe Foobar can also tell you the codec used to encode the MP3 (I am using a heavily modified version of Foobar, don’t remember the defaults.)
AllDay posted a very important picture too…the spectral analysis of a 192 kbps MP3.
If you (not you specifically, just in general) got a lot of your music from dodgy places, chances are you have a ton of lossy transcodes. The most common lossy transcode is someone taking a low bit rate file and converting it to 320 kbps. They think that if they do that, they’ll get better quality files. This is incorrect. Once you make an MP3, data is removed from it and can never be restored. When people convert a lower bit rate file to a higher one, they’re actually making the quality worse.
Spectral analysis can reveal these lossy transcodes. A proper 320 kbps MP3 will cut off at 20.5 kHz. If you’re looking at a file that says it’s 320 kbps but the spectral looks like the 192 kbps image AllDay posted…surprise! You have a lossy transcode and it’s pretty much garbage.
DJs and producers who edit MP3s in a program like Audacity and then save them again are also transcoding.
You might think you’re safe if you buy from sites like Beatport, but sometimes producers upload lossy transcodes too! Burial’s recent single “Temple Sleeper” was a lossy transcode. Warp Records even used a horribly encoded MP3 with artifacts on Aphex’s “26 Mixes For Cash.” So even CDs aren’t necessarily safe!
In general, you want to keep 320 kbps and V0 MP3s. 256-192 kbps is usually okay too. Some old net labels I like only ever released their music in 192 kbps, for example, and it sounds fine. I would definitely delete anything below 192 kbps though.
If in doubt, replace the track. Watch out for dodgy re-encodes (this is only an issue if you acquire your tracks blogaliciously) … As mentioned you can compare the file size to a legit HQ track of the same length or throw the tune in a spectrum analyzer.
P.S This process took Jester 3 years, all my tracks are now 320s
LAME is ubiquitous. But, for high(er) bitrates I find that the Fraunhofer mp3 encoder does a better job. The difference is “obvious” in a spectrgram when encoding from lossless. And I even think I can hear it…but that’s a hard thing to quantify. YMMV.
Below 192kbps, LAME makes some “hard” choices that tend to result in a better listening experience (low pass at 16kHz for instance)…but that’s all relative given the lower bitrate. For spoken word narration, podcasts, and even “sample” edits for clients to review I prefer LAME when working at low(er) bitrates.
I see the encoder in my Windows file browser. I may have a plugin loaded to enable that feature, I don’t recall.
In either case, I use MediaMonkey for all library maintenance, and MM will display the encoder used for mp3’s.
Lots of good advice in here…but all for “Step 2” of the sorting process. If an mp3 has a bitrate below 192kbps, you can directly sort it into the “LowBitrate” bucket and act accordingly. That tends to go a long way in any first time sort of a music collection.
This is wrong tho, because if a file is wrongly converted from say 192 upto 320. It will show as 320 in explorer. But the file itself isnt that quality.( you cant make the quality better from a compressed audio file )
That works for immediately spotting and removing low bit rates, but like AllDay said, the problem is if the file is a lossy transcode.
The original poster said:
That to me implies they downloaded their files from “unofficial sources.” There’s a huge probability of lossy transcodes with that, or bad encodes that have digital artifacts in them, which a lot of people don’t seem to hear, and of course “mutt rips”…
Spectral analysis is the only way to verify the actual quality.
This is true. But, it remains a “Step 2” concern. If a mp3 shows as 128kbps, that can be safely sorted into a “Low Bitrate” bucket. For DJs who have obtained their music in and orderly fashion, this is likely sufficient.
If a file has been transcoded to a higher bitrate, that is a harder nut to crack. There is no “automated” method to detect songs that have been treated in this manner. This kind of inspection is a VERY manual process. In practical terms, if I had concerns about the quality of an mp3 @ 320, I would get a new version before I would pull it up in a spectragram (but that’s me). I would update the mp3 by either converting it again myself from the FLAC rips I have done from CD, re-downloading the song from where I purchased it (if applicable), or repurchasing the song if it was that concerning to me.
By all means be careful when working with your music collection. But don’t drive yourself nuts.
