Quick question: is a FLAC file with the following media info still 'better quality' as a similar MP3 with a higher kb/s?

Acapella track > 1:19
v1) 231 kb/s - 44.1 kHz - 16 bits - 2 channels - FLAC
v2) 320 kb/s - 44.1 kHz - 16 bits - 2 channels - MPEG Audio

It's the bitrate I'm curious about. The shorter the FLAC track is, the lower this number. But what if it's even lower compared to a 320 kb/s MP3? Or am I missing the point that, despite this number, MP3 is always compressed with a certain loss of quality?

 

Check these tracks in program called Spek and send here screenshots from Spek. Or send tracks here

 

The FLAC audio track will reproduce the same PCM wave form that was input as long as sample rate and bit rate are the same. Although an 320kps 2ch (not joint stereo) mp3 will audibly give a FLAC file a run for its money it will not necessarily null with the source as a FLAC file will.

Just try it for yourself, use the uncompressed source and MP3/FLAC files and phase invert them and test them one at a time versus the source.

 

How did you get that low kbps value? When I export as FLAC it has 728kbps on average settings (in Reaper).

A 728kbps FLAC still nulls with a 1411kbps WAV.
But a 320kbps MP3 not, that's true.

 

do yourself a blind test with same song in diff format.. pick the one you like most and see what format it is.. :p

 

Probably not necessary. Hardly anyone can distinguish WAV and 320kbps MP3. A nil test is the way to go, IMHO.

 

okay.. nil test is probably new vocabulary for some of us here.. hehe.. how to do nil test @No Avenger ?

 

I'm aware of Spek but I prefer to use Fakin' The Funk (which has also an analyzer). Both results look the same though.

Unfortunately, I don't have the source material (CDM) so I had to download it. Otherwise, the problem solved itself by ripping it in WAV. But to be honest, I had the same thoughts! Although it occured to me that a lot of FLAC files with a short duration drop their bitrate. Normally that wouldn't be an issue but in this case the amount dropped beneath a 320 kb/s MP3 which makes it confusing.

It's only a vocal/speaking part. If it was music, comparing by ear would be easier yes. The full info of this particular MP3 is btw: MPEG Audio (Version 1) (Layer 3) (Joint Stereo / MS Stereo) so in this case that rules out the "not joint stereo" part. However, the first part seems to be the answer on my question! Also confirmed by No Avenger.

 

Oh, really? Aka nulltest like @JMOUTTON mentioned?

Just take two files and phase invert one. FLAC and WAV cancel each other out, 320kbps MP3 and WAV or FLAC not.

 

Last edited: Sep 10, 2020

Interesting. I tried it with a six seconds drumloop and the bitrate dropped down to 672kbps. Not even close to the 231 you mentioned.

 

Could it be that people choose a low encoding level? I mean, when I check the options inside dBpoweramp music converter (which I use myself) there are several options when it comes to FLAC. Lossless uncompressed but also level 0 (fast) to 8.

 

Reaper too has an option ranging from 0 (fastest = highest bitrate) to 8 (slowest = lowest bitrate) but I don't get it below 664kbps. Maybe this has also to do with the content (less freqs in speech vs full range in drums)?

 

60 seconds of silence encoded as a stereo mp3 @320kbps will result in a file of about 2.4MB.
60 seconds of silence encoded as FLAC will result in a file of 10kB size.

The mp3 encoder will maintain the constant bitrate no matter what you feed it while FLAC will remove all redundancy.
Got it?

 

Flac is always loseless, mp3 is not.
To compress the size of a file, flac format does not affect the audio quality (as happens with mp3): flac compression acts on the data, exactly as it happens with a zip file. In the flac settings there is a slider that allows us to tweak compression level (from 0 to 8) it modifies the final size and therefore the encoding duration, but not the audio quality. decoding process is always quite fast and not dependent on the level of compression.

@JMOUTTON The reversed phase test is interesting, I never thought about it, it makes sense.


@fiction

I did test what you say and it is absolutely true, first time I notice that

 

Last edited by a moderator: Sep 10, 2020

To dig a bit deeper in this subject. When I convert a FLAC file to WAV the result shows us 1411,2 kb/s. Now it's uncompressed but the audio quality should be exactly the same, right? Btw, a WAV file could be - according to MediaInfo - PCM (Little / Signed) and PCM (Big) but I'm not sure what's the difference. Same goes with extracting audio from video: DTS or DTS XLL (DTS-HD Master Audio) or DTS (Header Stripping). Although I do think that's the difference between 16 and 24 bits. Talking about 24 bits. That seems to be 2117 kb/s.

 

Sampling freq. in (k)Hz × bit depth × channels = bitrate in (k)bit/s

E.g. for CD: 44.1 kHz × 16 bits × 2 channels = 1411.2 kbit/s

The endianness (big/little) defines the byte order of the file. It can affect compatibility in some cases, but not the audio quality itself.

And don't forget to sample sync the (decoded) MP3 file, as MP3 is not sample accurate.

 

Last edited: Sep 10, 2020

Constant bitrate compression is a bitch

But yeah, it's like someone said. FLAC is like compressing a wav/aif file into a zip or rar. Any lossless (untouched quality) compression has an arbitrary size depending of the diversity of the source.

Or a png image. If you compress to png a picture which is 95% black(that's like silence in audio) it'd beat a jpg.

 

Since also the levels are changed, I wonder if this is really possible.

 

Mp3 uses psychoacoustic filtering and spectral cutoff, reduces dynamic range and uses quantization of dynamic changes, all in order to reduce the file size. So it entails intentional irreversible loss of data (that is why it's called 'lossy compression'). It also introduces DC offset, stereo offset and some transient loss as a by-product.
FLAC does not do anything of the sort, it just temporarily removes redundancies which are restored during the playback.
So the question 'is this FLAC still better', in my opinion, does not deserve any extensive discussion.

 

an mp3 on 320kbps constant bit-rate ripped from a 16bit 44.1 kHz source, perception wise, moves around in a area border line between psychology and physical hearing. i can perceive "something", but i cant say its a deterioration of the sound, its just "something" and only sometimes .