FLAC 1.4 Released With AArch64 Optimizations, Faster x86_64 FMA

[binarybanana]

Opus is amazing. I encode to 96k (default bitrate) for listening on the go and it sounds great even on my DT770s. But it's not an archival quality format. It's not a 1:1 copy of the source material. If you're serious about your music that matters even if there is no audible difference.

Now, why do people care about this for music but not video? Even hardcore archivists usually don't bother with raw BD rips. Maybe because the source isn't lossless either, so the choice to reencode it into a more efficient format isn't as painful.

Whatever the case, long live FLAC!

[cl333r]

Opus is amazing. I encode to 96k (default bitrate) for listening on the go and it sounds great even on my DT770s. But it's not an archival quality format. It's not a 1:1 copy of the source material. If you're serious about your music that matters even if there is no audible difference.

I heard the human ear can't hear past 192K, from prev experience with MP3 I'd go with at least 128K.

Even hardcore archivists usually don't bother with raw BD rips. Maybe because the source isn't lossless either, so the choice to reencode it into a more efficient format isn't as painful.

I did a few 1080p BDRips to HEVC and I couldn't tell the difference.

[ehansin]

To add to the whole lossy vs. lossless discussion here, FLAC is pretty popular I think in the bootleg (authorized or otherwise) concert recordings community. People who are really into this are of course audiophiles, but I'd say it goes beyond perceived listening quality. For people who are into this, it is almost more about preserving the artifacts. I'll use Grateful Dead concert tape traders from back in the day as an example. Each time an analog cassette was copied from a copy, the sound quality degraded. In the era of digital preservation, a lossless format means the best representation of the the original can be preserved. In the minds and philosophy of those that are really into this scene, their take is along the lines of "Jimi Hendrix ain't coming back to redo that 1968 Winterland concert (or whatever), so we better preserve what we have."

Sure, encode your personal listening collection down to whatever you want, but those into this scene look down (to various degrees) on anything shared and distributed as lossy, for reasons explained. Just because you don't see a need for lossless does not invalidate the needs of others that do. Kind of like a lot of things in life...

[quikee]

Now, why do people care about this for music but not video? Even hardcore archivists usually don't bother with raw BD rips. Maybe because the source isn't lossless either, so the choice to reencode it into a more efficient format isn't as painful.

Well.. it's a big difference when a BD rip takes 25-50GB (and a uncompressed source over 200GB or more) vs. CD rip lossless Flac encoded that is around 0.5GB.

[Anux]

I heard the human ear can't hear past 192K, from prev experience with MP3 I'd go with at least 128K.

Whoever said that has absolutly no clue what he's talkin about. The human ear in general can't hear frequencys above 20 kHz but depending on age that value lowers to 16 (around age 30) or even 12 the older you get. Some babys might hear up to 21 kHz but it's hard to do tests with babys that can't communicate.
You can let a doctor do a hearing test to check your frequency range.
If you transcode music to 128 kbps with lame it shows you the frequency cut off at around 16 kHz and most people can't hear that high and then there are also no acustic music instruments that generate sound in that range loud enough to be heard anyway. The differences you will hear are the compression artifacts of MP3.

There is no exact MP3 bitrate where one can't hear a difference (thats what's call trancparency), it depends on the listener and on the source material. Harps for example are especially problematic in MP3 and are distinguished (by some) even at 320 kbps. If you give me 2 sound samples and one is 128 kbps MP3 I can consistently tell you which is which unless it's low commplexity sound. One hi hat please.

All those people that claim they can hear a difference in above 20 kHz music are either not compairing the same source material or only think they can hear it. No one was ever able to prove this scientific.

[binarybanana]

I heard the human ear can't hear past 192K, from prev experience with MP3 I'd go with at least 128K.

Note that I said it sounds great, not indistinguishable. Even 64k still sounds pretty OK, but the difference is quite obvious at a bitrate that low even without paying attention. It's just that the artifacts aren't jarring. 96k is close enough to transparent for casual listening for me that I don't get distracted by the quality. I guess that's why it's the default. To show off how low you can go at acceptable quality. YMMV, depending on your music, gear and ears, of course.

quikee
Right. Sounds plausible.

[Anux]

Even 64k still sounds pretty OK, but the difference is quite obvious at a bitrate that low even without paying attention.

Whoa, you either have really bad ears or bad equipment. If I listen to ~64 kbps MP3 i can't stand it, it's compression artifacts all over the place and audible frequencys are cut off.
Well if you're fine with it you can save a lot of space.

For me lame -V6 (~120 kbps) is the lowest i can listen to in loud environments but I much prefer -V4 where it approches transparency for me. At -V2 I can't hear any difference (maybe in some samples with hours of A/B testing). I'm a little spoiled because of many listening tests that made me sensible to MP3 compression artefacts. I usually listen to flac wherever I can but on my old smartphone I have only 12 GB for data therefore compromises have to be made.
With opus those thresholds are much lower, around ~96 kbps is allready better than lame -V5 (at least for me) and also my choice for space constrained listening. Opus at ~140 is totally transparent for me.

At the end it's personal choice what sounds acceptable and if you don't train your ear to hear the artifacts it's a big bonus on space savings. Also a tip, if you know your personal hearing threshold you can override the cut off frequency in lame and get fewer compression artifacts at certain bitrates (that should only be made if your the only listener ever, others might notice that instantly).

On a 1 TB drive you can store around 150 days of FLAC audio, that's allready more than 3000 CDs. I don't have that many CDs/Rips yet.

[cl333r]

Whoa, you either have really bad ears or bad equipment. If I listen to ~64 kbps MP3 i can't stand it, it's compression artifacts all over the place and audible frequencys are cut off.

IF he uses Opus instead of MP3 then it's not that bad, Opus is (far) better than MP3 at lower bit rates, IIRC Opus size/quality is starting to get better and better than MP3 starting at 192KB and going down. Opus is really 2 codecs: one for low and one for high bit rates, which is one of its key secrets why it's so good at both low and high bit rates.

[quikee]

Opus is really 2 codecs: one for low and one for high bit rates, which is one of its key secrets why it's so good at both low and high bit rates.

True, it uses SILK, CELT and a hybrid mode, but note that unless you go less than 32kbps only CELT will be used. SILK is optimized for speech at low bitrates and isn't that good with music and CELT is usually better until you lower the bitrate enough. That's why there is a hybrid mode where it balances between codecs in a certain bitrate range.

BTW.. I can't believe that Opus is already 10 years old.... oh my.

[Anux]

Apart from opus beeing the most efficient audio codec in all bitrates (exept from maybe 8kbps) it's also free software, low latency and fast to en/decode. It really is a master piece. I whish it would become as widely used as MP3 once was. It certainly has all ingredients but maybe MP3 is already good enough for the for the majority.