What lossy format should I convert my flacs to for portable listening?

You can try OGG or OPUS, they're good options for lossy audio.

I use GoneMad which supports both opus and vorbis.
So which one is better?
Is 128kbps vbr good enough or should I give it some more?
thanks by the way

Attached: 1454283347867.jpg (1000x1167, 209K)

the correct format is opus vbr at ~160k target, or mp3 v0

It's impossible for the human ear to perceive the difference between 128kbps mp3 and lossless. Anyone who says otherwise is lying or is willingly eating placebos.

Whats it like living with permanent hearing loss

Use Opus. 128kbps is plenty.

Attached: 1010px-Opus_quality_comparison_colorblind_compatible.svg.png (1010x750, 67K)

I'm writing myself a script for converting. this should be fine then?
ffmpeg -i input.flac -vn -y -c:a libopus -b:a 128K output.ogg

Attached: Ludwig van Beethoven.jpg (650x781, 141K)

Opus is efficient enough you don't need anywhere near 160k for stereo. Anything between 96k-128k VBR is the sweet spot IMO.

V0

Looks fine.

Android doesnt have an Opus hardware decoder. So expect your battery life to take a hit.
You want to wrap that statement in a for loop unless you want to transcode your whole library manually

>You want to wrap that statement in a for loop unless you want to transcode your whole library manually
I did, but didn't think it was necessary for rating the encoding parameters

I miss my Rockbox player. I miss my Opus.

AAC 256Kbps VBR

AAC is the replacement for MP3. Your device probably supports hardware decoding.

Wave diagrams saved as JPEG at 20% quality setting.

Good enough if you have to ask.

AAC

more like KYS

Rebbit spacing

More like

You have to go back

Is FLAC and other lossless codecs legitimately lossless or is it just a REALLY good approximation of true lossless?

I mean, you can always sample faster, use a higher bitrate, and higher bit depth?

Even though no software supports it at the moment, if I recorded music at 1MHz, 128-bit, and 1Gbps are you seriously telling me that would be equally as good a representation as 96kHz, 24-bit, and 1024kbps? It wouldn't be even just a little closer to the true audio, even if the difference is effectively inaudible?

Based on that, it's right to conclude that lossless isn't TRULY lossless.

That's not how it works

Okay, well how DOES it work?

If you sample something at X Hz, you'll be able to perfectly reproduce all signal components below X / 2 Hz. 100% perfectly, with no error, given sufficient bit depth. Humans can only perceive sound from around 10Hz to 20kHz.

This is why a standard audio CD is 44.1kHz and 16 bit depth. With dithering aka error diffusion, this can reproduce essentially any sound perceptible to a human.

What lossless means is that these numerical samples, whatever their sample rate and bit depth may be, are reproducible perfectly from the lossless audio file. Ie with lossless you can go back and forth between raw samples and lossless audio file without any errors being introduced, just like you can zip and unzip and archive repeatedly on a computer without messing up the file.

Lossy audio makes approximations of the raw audio samples in order to save space. So if you decode a lossy audio file, you don't get the exact same raw samples you started with - instead you get something different, but made to sound as close as possible to a human. Lossy audio is like JPG image compression.

MP3 V0. Supported everywhere, sounds fine, small enough.

Opus VBR if you have a specific application you'll be using to play the files that you know supports Opus.

It has nothing to do with conversion. If you reduce the amount of information in the data, of course you're going to lose information.

If you're trying to squeeze every last bit of space, sure, but I am with the other user and agree that 160 kbps is better if only to throw extra bits at trouble samples.

opus

Thanks for the real answer.

My one comment which is more tangentially related though I'll mention it since you brought it up... I find the Nyquist sampling theorem woefully insufficient. You absolutely cannot accurately reconstruct a signal using a minimum sample rate of twice the highest frequency. Like, you cannot reconstruct on period of a sinewave with just two samples. Depending on where you sample you'll get different results and none of them will be accurate. Worst case sample at 0 and π and you get an zero. Literally no signal since that's the zero crossing point of the waveform. Sample at π/2 and 3π/2 and you get a positive and negative impulse at the peak amplitude of the waveform. Even if you low pass filter it you will only get a sawtooth waveform out. Still nowhere close to a sinewave. Ideally I'd say the sample rate should be 10x the highest frequency though even that would probably still give you a heavily distorted sine wave that would require filtering.
I'd imagine the only way a lot of audio got away with the 44.1kHz sample rate for is because it's only creating distortion in the higher frequencies that people are less sensitive to.

Serves you right for buying a phone without a micro SD card slot in 2018.

It's assuming the samples are evenly spaced, which they almost always are. If that's assumed, then you can unequivocally reproduce a signal using a sample rate of twice the highest frequency component.

Ogg Vorbis because 48khz is a meme

I still don't see how that's possible without at least low pass filtering to remove the harmonics.

Not him but you always low pass filter out anything above the nyquist frequency to prevent aliasing

Wav

Using outdated hardware in 2018

I would think the filter would not be necessary if the sample rate >>> highest frequency of interest. Sampling 20kHz audio at 20MHz should give you a very good reproduction, no filter required. Maybe not so great for high speed DACs in test and measurement equipment as it'd require microwave frequencies but for audio its fine.

Every time you listen to a song on your physical hard drive, it loses bits and eventually it will lose sound quality. That is why the only proper form of listening to music is through Youtube where the music is stored in the cloud.

You don't really want to capture the frequencies above 20kHz in the first place. It's unnecessary information to encode that just makes everything harder later, like the effectiveness of dithering down to 16 bit samples for the final master.

Yeah sorry, I should have stated it correctly the first time. I was right in but was subtly incorrect in

I copied my post from identical thread the other day.

20yr audio engineering veteran reporting in.

> shittest format for sound and size tier
320kbps CBR MP3 (literally half the size of a flac
> best just werks lossy tier
MP4 / AAC @ 192
> autismo tiny size lossy tier
Opus @ 96
> best muh redbook 16bit just werks tier
Flac
> muh annoying tier
Ape
> muh hires 32bit PCMlet with metadata tier
Wavepack
> muh undisputed king of the world archival 2inch 30ips reel to reel tier
DSD 5.6mhz rips from high quality analogue sources ... And if you haven't heard native chad DSD at 5.6mhz or 11.2mhz then you are simply a girly man and have no business being in this thread... /thread/

ogg >>>>>>>>>>>>>>>>>>>>>>> mp3

AAC produces weird artefacts on a way higher number of test samples than MP3.

What the fucking fuck man. MPeg 1 layer three is a dinosaur format designed in the 80's for Movie soundtracks and its shit

MPEG 4 AAC is much better as it was designed in the 90s and if you want engineering specifics it has a variable window size on its discrete cosine transform window size which makes it sound better and more efficient while MP3 is fixed. So shut your pie hole.

that's utter bullshit. I can easily distinguish FLAC from 128 kbps.

However, it's really hard to select which one is better between 320kbps and FLAC

You can ABX test yourself with foobar if you want to determine what the best you can distinguish is.

Midi

Attached: 1486354482016.jpg (200x232, 17K)

>thread about codecs
>OGG container format
>"I only get MKV movies because they have the best quality"

Attached: 39494352322.png (640x985, 164K)

>flac autist asks if 128 kb/s is enough instead of just trying because he knows he won't be able to tell difference between a youtube rip and flac on his shitty earbuds anyway

He probably meant Vorbis.

Doesn't this image only imply that the sample rate is higher

let's see..
>320kbps CBR MP3 (literally half the size of a flac
mp3 @ 320 kbps cbr offers no sound quality advantage over mp3 V0 and the latter takes much less space
320cbr is just wasting storage space
>Opus @ 96
Opus is lossless at 64 kbps already, but of course it depends on the sample and 96 would indeed be a nice all-around bitrate to bump the quality a bit for harder samples
>DSD
a meme, you wouldn't be able to tell the difference in an abx test

i meant transparent, not lossless when talking about Opus

mp3 v0 is actually transparent.

>talks about mp3 320
>no mention of v0
more like 20yo

First supported by whatever you want to play it on:

1. Opus at 128k
2. AAC at 192k
3. MP3 v0

this is reasonable

This. It's too good to miss out on.
auphonic.com/blog/2018/06/01/codec2-podcast-on-floppy-disk/

You should try opusenc, too, and compare the performance.

> Opus @ 96
> Opus is lossless at 64 kbps already
are you guys for real? ive been under acrock and this happened? convince me to switch from v0, cause I just don't believe any format could sound good at 128, not even gonna look at stuff below that

Nailed it

AAC 256 VBR, you will not hear any difference. Also this is an iTunes standard for many years.

just test it yourself
foobar2000 has a convenient ABX Comparator plugin which lets you do a blind test comparing two audio files
just make sure you create the opus encodings from a confirmed lossless source, because obviously you're gonna get skewed results if your source is a fake flac or a legitimate lossy master
try it with opusenc.exe --bitrate 64 or opusenc.exe --bitrate 64 --framesize 60

Just wait™ for opus 1.3 stable

iTunes standards suck desu

mp3 is just a container

You're technically correct, as higher sampling frequency will further apart the spectral replicas thus eliminating the aliasing, so 1 MHz would produce a finer sample than 1 kHz.
It seems fairly obvious however that more samples = more data to be stored and flac by themselves aren't quite small and we're talking around 100 kHz sampling rate, if you put the frequency on steroids however you could easily end up with a 1 gb per song. There's always a trade-off, in this case quality vs space requirement & hardware cost

fucking lol
i got around 1,5 GB of free space in my phone and still listening to flac
i may be peasant, but at least i try

sauce

A good phone with enough storage.

Spotify

>downloading the song once
vs
>downloading the song everytime
don't know about you but for me mobile data is >> expensive than my connection at home.

They literally do not add/count it up to my plan

yes, it's for brainlets that don't understand sampling theorem

isn't that for voice only?

summerfag please go

/thread

Clean your ears, faggot.

Opus, it's vastly superior to anything else.

/thread

Attached: FLAC_MP3VBR.png (2528x614, 1.82M)

Correct. Also, the human eye cannot see more than 25 fps.

>66899709
BRRRRRRAP

Only if you're a pussy.

320Kbps MP3 is not that bad but it too suffers from bitrot

CBR algorithm is no longer improved. You should use VBR.

Are you really a girl? Can you show feet as a proof?

I'm a woman. You can kiss my ass for your "proof".

Or even better, use find with exec.

LARPing faggot

Coming from an incel it surely means a lot.

can you, I mean movies are still produces at 24 fps and that's supposed to be the professionals