The Lossless Video Game Soundtrack [Discussion] Thread

[renikrill]

Counterpart to: [Hidden link. Register to see links.]

Fill with your assortments of jibble-jabble & the like; informative and/or argumentative, all with regards to the quality representation (or misrepresentation, if you like) of lossless audio.
Every bit of that sort of talk goes in this thread and NOT the above one (which is for d/l links and specific talk with regards to the links/files themselves...)

Peace :-)

---------- Post added at 11:36 PM ---------- Previous post was at 10:52 PM ----------

Could you help me out in describing what subjectively sounds different about the lossy encodes below? I'm curious to know about it. My high frequency hearing is terrible, but changes in other frequencies I can hear for myself when they are obvious. Or just rank how good you think they sound from best to worst. One of them is the lossless, while the other three are different lossy bitrates/formats.

Folder link:
[Hidden link. Register to see links.]

Individual links:
[Hidden link. Register to see links.], [Hidden link. Register to see links.], [Hidden link. Register to see links.], [Hidden link. Register to see links.]


The music is from the same lossless file, "The Shadow of Numeros" from The Legend of Heroes V - A Cagesong of the Ocean (falcom jdk). If anybody wants the whole album (ape+cue, not my rip), let me know!

[Hidden link. Register to see links.]
[Hidden link. Register to see links.]

It's nigh impossible to "describe" specific differences as you hear them in any given track. The way I see it is you can either hear the difference, or you cannot. Simple as that (engineering technicalities aside).
It's not easy to pick out a quality difference given a track you've never heard before, especially one with little dynamic range, so I'd be completely unfazed if I'm incorrect in guessing the following:

alpha and delta are supposedly mp3, or something of a lower caliber (but I'd bet on mp3) 'delta' is the higher bit-rate of these two.
gamma must be the lossless encoding (this seems the crispest of the bunch), and beta is something of a considerably high-fidelity (I'd put my money on AAC?)

I cross-compared alpha and delta on a spectrograph, because I couldn't tell the difference in clarity between them from listening alone. They were obviously the 2 lowest-quality though.

[Albanoid]

[Hidden link. Register to see links.]

also would be prudent to know if anyone testing zeaot's files is doing ABX tests in fb2k and mentioning how many times they're doing it.

[zeaot]

Oh, I forgot this thread was made. I'm sure nobody cares which was what, but for completeness:

What is what is in the spoiler below (I think--I didn't bother saving these and am going off of memory and others' descriptions. I recall that Arceles did a pretty good job figuring them out, so reading that response jogged my memory)

Spoiler!

alpha - LAME 3.98 mp3 -V0 (was like 250 kbps average or so)
beta - aoTuV Vorbis 6.02 -q8 (ended up like 280 kbps average or so)
gamma - original lossless
delta - aoTuV Vorbis 6.02 -q5 (was like 170 kbps average or so)

For portable use I transcode lossless to my "beta" aka latest Vorbis -q8, which is pretty overkill for most music. Maybe it's better to use something lower so I can fit in more tracks, but certainly it works out better than storing FLACs on microSD cards. Uh well, version 6.03 is out now, but that's a bugfix for 5.1 encoding.

Anyway, one point is maybe that mp3 sucks aside from ubiquitous interoperability, and maybe lower complexity for encoding/decoding. There have been many other listening tests on this subject though, so this is nothing new.

Also maybe it's interesting to note that renikrill and Arceles had a different opinion of which was worst.

Personally after ABX I think all of them sound fine, just that there are indeed places where if I crank up the volume and compare carefully, I can tell a difference between the two worst ones compared to either of the two better ones. The two worst ones also sound different in a few places and screw up the music in different ways, but it's mostly subtle. I can't tell between the two better ones at all, but other comparisons are like maybe 8/10 ABX after some training. I didn't sit down and test all day though.

I think the slide up near the beginning and some of the percussive effects are not very kind to most encoders, though this is not a particularly special "problem" track. But I know I've tried to ABX mp3 encoded with LAME 3.98 -V0 with lossless and have not been able to do so, on most music. Here I was able to barely.

[Daytime Dreamer]

I've seen noted in the lossless audio that some rips that are suppose to be FLAC rips are not. Some people check those with a program that displays a waveform. Which program is that and can anybody put up a mini tutorial how to check FLAC files to be true FLAC indeed?

[unrealmaster287]

strawman ,
the only reason i use WMA lossless 9.2 !! is because it is the only format with fyll windows property handler support as well as WMP zune and other MS products

I think that chosing a music player comes down to personal preference i personally use WMP with shark007 but if anyone likes th foobar interface then they are provably wrong :P

Ok i am a microsoft die-hard user, so i use WMP!
Get over it
Oh and thank you so much for the cuetools link

But the thing that worries me about those is that am i guaranteeed lossless in lossless out, or lossless in bad pecie of **** out, if you know what i mean

that is why i am very picky about chosing my audio tools

[LiquidAcid]

Ok i am a microsoft die-hard user, so i use WMP!

that is why i am very picky about chosing my audio tools

You're contradicting yourself.

[unrealmaster287]

You're contradicting yourself.

lol
Nice one

i meant audio converters/splitters

[alc123]

if anyone likes th foobar interface then they are provably wrong :P

Yawn.

But the thing that worries me about those is that am i guaranteeed lossless in lossless out, or lossless in bad pecie of **** out, if you know what i mean

CueTools doesn't alter the audio data itself in any way. It's simply a tool for changing formats.

[unrealmaster287]

hey guys i took your advise and tried cuetools but i have a number of questions:
When i chose the output as wav no tags are saved (why)

Second when i chose flac a drop down spears containing

• libflac
• libflake
• libflaccl
• flake

and i also get a slider bar with numbers the range depends on the option

to get the absolute best quality which one should i use and what number :P

OK reading back that sentence felt like i was asking a question on those game shows :P

Thanks in advance

[strawman]

if anyone likes th foobar interface then they are provably wrong :P

I love foobar, but I do wish that it looked more like a Taiwanese overclocking tool.



Fortunately, while I can't make foobar look that awesome, I did hack in a custom menu which lets me set "FSB" to "COLONEL" and then immediately causes a bluescreen, so I'm halfway there.

hey guys i took your advise and tried cuetools but i have a number of questions:

You can't tag WAV files, the format doesn't support it. There's no reason to export to WAV, just go directly to FLAC.

Also, all of those FLAC encoders should theoretically give identical quality, since FLAC is lossless and all; the only difference is encoding time and filesize. If you have a GPU that supports CUDA/OpenCL then use FLACCL, otherwise just use libFLAC. Whichever one you choose, set it to level 8 unless it encodes really slowly. Levels 9-11 for FLACCL aren't officially supported, so some FLAC players might not be able to decode them.

[unrealmaster287]

cue tools help



hey guys i took your advise and tried cuetools but i have a number of questions:
When i chose the output as wav no tags are saved (why)

Second when i chose flac a drop down spears containing

• libflac
• libflake
• libflaccl
• flake

and i also get a slider bar with numbers the range depends on the option

to get the absolute best quality which one should i use and what number :P

OK reading back that sentence felt like i was asking a question on those game shows :P

Thanks in advance

[alc123]

to get the absolute best quality which one should i use and what number :P

That you're asking this indicates that you don't understand what "lossless" actually means. If there was any difference in the audio quality between any of those FLAC implementations then it wouldn't be lossless, because a change would've occurred.

All of your other questions have already been answered in [Hidden link. Register to see links.] post. Duplicate posting is probably discouraged here, btw.

[unrealmaster287]

That you're asking this indicates that you don't understand what "lossless" actually means. If there was any difference in the audio quality between any of those FLAC implementations then it wouldn't be lossless, because a change would've occurred.

All of your other questions have already been answered in [Hidden link. Register to see links.] post. Duplicate posting is probably discouraged here, btw.

About the duplicating posts it was unintentional i had two tabs open on the same page, one i used to write the post and the other to brows, so when i finished writing the post i was studying, when i came back i looked at the other open tab it did not have my post on it (forgot to refresh honest) and so thought it was a bug (it had had happed to me before on dbpoweramp forums by the way) so i retyped the post again, well not really i save all posts on a notepad (in case i sped 10 mins writing a post and it does not appear)

Believe me or not honestly it was an mistake

second i do know what lossless is i know that wav will give the same quality as flac etc (be easy on me i only got into this lossless scene last month ) i was just wondering do the numbers indicate something to do with altering the quality or something, later after ABOUT 25 mins of searching i found out that the options are different encoders and the numbers represent compression.

Which i still dont understand what does compression affect other than file size and if the answer is nothing why dont we compress all our flacs to 8

So there you have it

[Leon Scott Kennedy]

Which i still dont understand what does compression affect other than file size and if the answer is nothing why dont we compress all our flacs to 8

Encoding/Decoding speed, filesizes (although these aren't always THAT noticeable). Regarding the last question, the answer lies in one of the most obvious reasons: most users simply don't care about these settings and leave the default compression level (usually 5/6, depending on the software).

[alc123]

Encoding/Decoding speed

Honest question, isn't the decoding basically speed the same regardless of FLAC compression level? I was fairly certain higher levels didn't add a significant computational hike, but I might be wrong.

Historically I think the reason for choosing something other than FLAC lvl 8 is encoding time, which was at best a tenuous argument five years ago (spend a few minutes, save a few megabytes, why not?), but frankly anyone whose computer is genuinely struggling with encoding to lvl 8 in this day and age needs to get with the program and upgrade to something that isn't made of steam and clockwork.

Admit it, you secretly wish it looked like this:

I tend to forget that software like that actually exists.

Ick.

[LiquidAcid]

@unrealmaster287: Read the section about "Software vs. format, format vs. codec" on [Hidden link. Register to see links.]. FLAC defines a bitstream format and also a container for the stream.

[tangotreats]

Repeat after me:

1) Lossless is lossless. The only variable metric is the file size.

2) Re-encoding lossless format X to lossless format Y yields the same sound quality. It is not "theoretically" the same - it is the same.

3) Decoding to WAV first is a waste of time - it destroys tags, takes longer, and does nothing for the sound quality. The converter is decompressing the audio prior to re-encoding it anyway.

4) Higher compression takes longer to COMPRESS but all about the same to decompress (with the exception of APE at the highest compression level - which is fairly intensive to decode). For the most part, encoding is the mathematically difficult part. Decoding is easy. In 2011 there is no reason whatsoever to use anything but the highest compression level. If the user at the other end wants something different, he can re-encode it himself without quality loss until his heart is content. Lossless is diplomatic; it may not be what EVERYBODY wants but it can be turned INTO what EVERYBODY wants easily by the recipient. Save bitrate and use the highest compression level. It's quicker for you to upload and quicker for the recipient to download.

4) Foobar2000 is fantastic. If you don't like the interface, it can be customised. It is probably the best player out there and most people's converter of choice.

5) Aren't those overclock things UGLY! I suppose it's to distract you from the fact that they basically do nothing.

[LiquidAcid]

4) Higher compression takes longer to COMPRESS but all about the same to decompress (with the exception of APE at the highest compression level - which is fairly intensive to decode). For the most part, encoding is the mathematically difficult part. Decoding is easy. In 2011 there is no reason whatsoever to use anything but the highest compression level. If the user at the other end wants something different, he can re-encode it himself without quality loss until his heart is content. Lossless is diplomatic; it may not be what EVERYBODY wants but it can be turned INTO what EVERYBODY wants easily by the recipient. Save bitrate and use the highest compression level. It's quicker for you to upload and quicker for the recipient to download.

This can be even extended. The major bottlenecks on most modern systems (both desktop and portables) are storage and memory bandwidth. Even if the computational complexity increases a bit with higher encoding efficiency, this is by far outweighted by the bandwidth benefits. This of course doesn't apply to symmetric codecs.

[alc123]

It is not "theoretically" the same - it is the same.

Agreed, though I think when strawman said "theoretically" he meant "if everything's working correctly" (see also: Medieval Cue Splitter's habit of altering the audio stream).

[unrealmaster287]

Repeat after me:

1) Lossless is lossless. The only variable metric is the file size.

2) Re-encoding lossless format X to lossless format Y yields the same sound quality. It is not "theoretically" the same - it is the same.

3) Decoding to WAV first is a waste of time - it destroys tags, takes longer, and does nothing for the sound quality. The converter is decompressing the audio prior to re-encoding it anyway.

4) Higher compression takes longer to COMPRESS but all about the same to decompress (with the exception of APE at the highest compression level - which is fairly intensive to decode). For the most part, encoding is the mathematically difficult part. Decoding is easy. In 2011 there is no reason whatsoever to use anything but the highest compression level. If the user at the other end wants something different, he can re-encode it himself without quality loss until his heart is content. Lossless is diplomatic; it may not be what EVERYBODY wants but it can be turned INTO what EVERYBODY wants easily by the recipient. Save bitrate and use the highest compression level. It's quicker for you to upload and quicker for the recipient to download.

4) Foobar2000 is fantastic. If you don't like the interface, it can be customised. It is probably the best player out there and most people's converter of choice.

5) Aren't those overclock things UGLY! I suppose it's to distract you from the fact that they basically do nothing.

Ok miss,i have learn my lesson, now may i please go to the playground and play with all the other kids :P

Sorry i just had to say that, anyway thanks for the info, i now fully understand the differences between lossless formats (there aren't any) exept for file sizes

Second could you show me a screen shot of a decent foobar setup (theme) that looks normal

And please i am not a pc noob (just when it comes to digital audio), i mean i know how to program (c#,C++ and VB.net), but when it comes to digital audio i know very little yet i want everything i dont care about size i dont care about format (thats a lie i do care about the format) i just want to sleep knowing that i have the best sounding audio i can get for my money (come to think of it i spent $0 so lol) I just want a ..............

Wait a second if lossless is lossless then why is there WV,WAV,WMA,FLAC,etc formats O.o
Relax i learn my lesson but honestly why so many formats


PS:
Thanks to all those that helped in my learning process ( i wont mention names because they will fill the page),
I apologize for any 'bad'/dumb questions i posted, and for that double post (Which isn't my fault ) but still

So thank you in advance

Good night ladies and gentelmen

Music fades in
-----*CREDITS*-----
Text - me
Questions - me
Answers - Everyone else
Unrealmaster287 played by himself
Every one else played by Them (points at you, yes especially you Michel)

can't blame me for having some fun

[LiquidAcid]

i mean i know how to program (c#,C++ and VB.net)

*rofl*
Based on your comments and questions here I doubt you could code your way out of a paperbag...

[tangotreats]

This can be even extended. The major bottlenecks on most modern systems (both desktop and portables) are storage and memory bandwidth. Even if the computational complexity increases a bit with higher encoding efficiency, this is by far outweighted by the bandwidth benefits. This of course doesn't apply to symmetric codecs.

Quite right, didn't think of that! Modern systems can probably decode FLAC faster than the storage medium can physically provide the raw data.

Agreed, though I think when strawman said "theoretically" he meant "if everything's working correctly" (see also: Medieval Cue Splitter's habit of altering the audio stream).

Well, something pretty bad would have to be going on for the audio to be erroneously encoded to a lossy format and decompressed somewhere in the chain... but point taken. It's important to clarify though that there is no "maybe" or "theoretical" as far as the actual process goes. If you do something silly, or your encoding software is doing something silly, anything is possible...

Ok miss,i have learn my lesson, now may i please go to the playground and play with all the other kids :P

Sorry i just had to say that, anyway thanks for the info, i now fully understand the differences between lossless formats (there aren't any) exept for file sizes

Second could you show me a screen shot of a decent foobar setup (theme) that looks normal

I didn't mean to be patronising; I just wanted to collect a few facts together in one place.

Whether a Foobar setup is "decent" to you or not depends entirely on what you want out of it. It's pretty crummy right out of the box but if you spend a little time on it, learn how the customisation works (it's hard work, but it really is worth it) you are virtually able to build your own player from scratch.

Wait a second if lossless is lossless then why is there WV,WAV,WMA,FLAC,etc formats O.o
Relax i learn my lesson but honestly why so many formats

There's always more than one way to do essentially the same thing. All formats are working to the same goal but use different mathematical tricks to get there. There's taking advantage of redundancy (ie, if you have a passage of data that's fifty zeroes, twelve ones, a hundred and fifty zeroes, you can store that in less space by saying 50x0, 12x1, 150x0 etc rather than by explicitly writing out 00000000000000000000000000000000000011111111111000 00 etc) which ZIP, RAR, etc all do as a matter of course. Lossless codecs go further; I do not understand the precise detail of how they work but I do trust that they do work. Lossless audio coding is probably the only audio-technology related piece of processing that you can prove worked without getting into the murky waters of subjective opinion and double blind testing. If you have some raw sound, you compress it, decompress it, compare it to the original, and the data stream is the same, it worked. If they're not, it didn't.

[LiquidAcid]

Lossless codecs go further; I do not understand the precise detail of how they work but I do trust that they do work.

It's not really that complicated. E.g. FLAC has basically five steps: blocking, inter-channel decorrelation, modeling, residual coding and framing

Blocking is kind of self-explanatory, you just hack the input stream into blocks of a certain length (VBS = variable block size is possible).
Inter-channel decorrelation is what's usually known as joint-stereo in the lossy/MP3 world. It's just taking advantage that e.g. the two channels of a stereo channel contain "mostly" the same signal.
Modeling now tries to approximate the signal of a give block with an "easy" mathematical function. Easy in the sense that you need only a few bits to store the type of the function. Now since this is only an approximation, you need the next step.
Residual coding now takes the difference signal (difference between approximated and input signal) and applies a normal data compression to it (Rice-coding in the case of FLAC). If the approximating function was well chosen by the encoder, than this step is mostly going to encode zeros (zero = no difference between approximated and input signal).
Framing is then just packaging of the computed data (the appromation function plus the encoded difference signal) together with some checksum and a header.

Well, and that's it.

[unrealmaster287]

Ok thanks all for all the help

And LiquidAcid, yes i know my questions were extremely dumb pointless and redundant but come on its a holiday here, i have nothing better to do than sit on a text editor and write some verbal diarrhea without thinking about it

And i do admit that when i posted that question for the different flac encoders and the compression bar thing, the fact that it is still lossless skimped my mind i thought it affected bit rate/freq/something

tangotreats, and LiquidAcid, and every one else thanks for the info,

PS:
I use VS 2010 SP1 to program in VB and C#
My VB.net skills are mid/expert
My C# is so,so
I use a program called kiss for my C++ but it is basically rubbish (i use it for the bot ball competition so i only need to use loops and pre-defined functions, sometimes i create my own functions )

Could we close this thread and forget it ever happened

[B l u e]

Great topic, but a lot of osts links are down.