"Sample Rate" -- extremely important, please read

Peter Why · Post by **Peter Why** » February 21st, 2006, 12:25 pm

Raouf,
With regard to my checking sample rates on the sound files, I wanted to do it without loading the files for playing (twenty poems! ... that I've already listened to once, as a quality check, mostly before this sample rate business came up). The file properties was quite convenient for this.
But thanks for the suggestion.

I'll just have to remember to check rate when I play them in future ..

Peter

kayray · Post by **kayray** » February 21st, 2006, 1:42 pm

LibraryLady wrote:I'm confused, I thought the problem was with mp3 players, not the validator?? Everything was fine and the validator gave us no trouble, I thought this all came up become someone couldn't play something on their mp3 player. Maybe I'm thinking of something else?

A little from column A, a little from column B :)

A sample rate of 48000 breaks the validator AND archive.org AND some mp3 players.

22050 and 32000 do not break the validator. We're fairly sure they don't break archive.org. I'm still waiting to hear back from the guy who reported the problem in the first place whether or not 22050 and 32000 break his mp3 player -- i suspect they will be fine.

In the interest of trying to confuse as few people as possible, we've declared 44100 our official sample rate -- it's definitely not going to break ANYTHING. People who record at 22050 or 32000 are not really a problem... we're just trying to keep things simple :)

kri · Post by **kri** » February 21st, 2006, 1:44 pm

kayray wrote: In the interest of trying to confuse as few people as possible, we've declared 44100 our official sample rate -- it's definitely not going to break ANYTHING. People who record at 22050 or 32000 are not really a problem... we're just trying to keep things simple

I agree. Even though there are other sample rates that might work, instead of saying "This may work, this does this, this shouldn't be a problem." It's easier when dealing with large amounts of people and trying to mass information to just say "This is how it is." Big period.

LibraryLady · Post by **LibraryLady** » February 21st, 2006, 2:29 pm

Okay, so the concensus is, we say 44100 is the standard, but if we get 22050 or 32000 it's not a big deal. Right?

HerrSchildkroete · February 22nd, 2006, 1:01 am

Hi all,

I have exchanged some private messages with linton and done some more research on his claim that for voice 32kHz will generate better quality MP3s than 44.1kHz and there is a lot of evidence to support that. In other words: that makes perfect sense to me.

Therefore I suggest to actually adopt the 32kHz, i.e. ask everyone to record in 32kHz and continue accepting 44.1kHz as well.

This is why: Technically using 32kHz is better because of some oddities in the MP3 file format. However at 128kbps the difference is barely audible. So we shouldn't stick knowlingly with the inferior quality. However we should not make a fuss about it, if people hand in 44k. Its definitely more important to have the recording than to have "perfect" quality.

I will update the WiKi to reflect my suggestion.

Regards, Felix

Gesine · Post by **Gesine** » February 22nd, 2006, 2:14 am

Hi Felix (and Linton), thanks for testing.

It'll be easier for people if we leave the LibriVox default at 44100, since two of our most widely used programmes (Audacity and GarageBand) have that as a default. Changing the rate to 32,000 will require action from users and confuse many (if we put it in our instructions, people will want to get it right). It's just one more technical thing to arrange before people can start recording.

In other words, if we decide to change the LV policy I think we need to have a guide on how to change the sample rate in most known audio software, and the BCs will need to link to that guide when they refer to sample rate requirements in their threads.

In any case, I'd like to see some consensus and general conclusions in this thread, before we make any more changes. What's needed is one final LV policy, and the wiki updated accordingly. Then this thread could be unstickied (it's far too long and confusing for a sticky, with conflicting information) and the decision incorporated in existing Stickies in the normal way.

HerrSchildkroete · February 22nd, 2006, 2:58 am

Hi Gesine,

Changing the rate to 32,000 will require action from users and confuse many

I am aware of that. That's why I proposed accepting 44.1kHz as well

(if we put it in our instructions, people will want to get it right).

But I was not aware of that - although you are certainly right. A solution here would be to ask for any of { 44.1k, 32k },without setting a preference and set up a page like "Advanced fine tuning" that explains the advantages of using 32kHz and other stuff like optimal recording volume settings, level of noise reduction, maybe how it works and so on. I have noticed that a lot of people are actually eager to learn and understand what they are doing when editing audio

Who will finally decide about the official policy? Right now on the WiKi page there is a note saying that the information on the page is still subject to change. That should be removed, once a decision has been made. From what I have seen so far, I suggest that tis should finally decide about that. Otherwise there are too many people and too many opinions around

Best wishes, Felix

kri · Post by **kri** » February 22nd, 2006, 7:34 am

HerrSchildkroete wrote:Hi Gesine,

Changing the rate to 32,000 will require action from users and confuse many
I am aware of that. That's why I proposed accepting 44.1kHz as well

(if we put it in our instructions, people will want to get it right).
But I was not aware of that - although you are certainly right. A solution here would be to ask for any of { 44.1k, 32k },without setting a preference and set up a page like "Advanced fine tuning" that explains the advantages of using 32kHz and other stuff like optimal recording volume settings, level of noise reduction, maybe how it works and so on. I have noticed that a lot of people are actually eager to learn and understand what they are doing when editing audio

Who will finally decide about the official policy? Right now on the WiKi page there is a note saying that the information on the page is still subject to change. That should be removed, once a decision has been made. From what I have seen so far, I suggest that tis should finally decide about that. Otherwise there are too many people and too many opinions around

Best wishes, Felix

I'm with Gesine on this one. 44100, because it's a common default, and we need to have one answer to this question. We can easily let things slide if people aren't adhering to this strictly, but it's much harder to deal with the confusion that is "This is what we want, but this might be OK, maybe, kinda...". If the quality difference is very small between using 44100 and 32000, we should just do what is the easiest. The fewer instructions people have to read (for example, how to change your default sample rate from 44100 to 32000) the better. We want to encourage people to volunteer for LibriVox, not discourage them with a huge instructions manual before they start

Linton · Post by **Linton** » February 22nd, 2006, 11:07 am

May I suggest being as simple and flexible as possible, but making an order of preference clear?

For example we could say:

1. 44.1KHz is the default sample rate in lots of software. We like getting files at this rate.
2. Some know-it-all

told us that 32KHz sounds better. If you know how, try to use this rate.
3. We want to make this easy, so we'll take any rate from 16 to 44.1KHz, if you need us to.
4. Please don't use 48KHz or anything under 16KHz. If you accidentally send us something at these rates, our uploaders will resample it for you and advise you on how to record at a sample rate we can use. Don't worry, your work will never be wasted!

I know this is a lot longer than saying "please use X KHz", but it's also a lot friendlier. While some technophobes might like a firm guiding hand, people who have a 'favourite' rate or can't use certain rates for technical reasons might be put off if we're too strict.

kayray · Post by **kayray** » February 22nd, 2006, 11:10 am

I really like that, Linton! I think it makes our wishes clear and isn't confusing or unreasonably demanding :)

Could we make 22050 the low end, though? I haven't done any testing of 16kHz in the validator or archive.org and I'm not sure it'll work.

I think Linton's list should be added to the appropriate place in the wiki.

vee · Post by **vee** » February 22nd, 2006, 1:13 pm

I would definitely avoid setting the low-end to 16. I'm tempted to say set the low end at 32, but I'm just being picky. I think a range of 22khz to 44.1 khz should be fine. We won't necessarily indicate a preference, other than the range. If we get the occasional 48khz, it's easy enough to downsample.

kayray · Post by **kayray** » February 22nd, 2006, 1:21 pm

Ok, how about this:

1. 44.1KHz is the default sample rate in lots of software. We like getting files at this rate.
2. Some know-it-all ;-) told us that 32KHz sounds better. If you know how, try to use this rate.
3. We want to make this easy, so we'll take any rate from 22 to 44.1KHz, if you need us to.
4. Please don't use 48KHz or anything under 22KHz. If you accidentally send us something at these rates, our uploaders will resample it for you and advise you on how to record at a sample rate we can use. Don't worry, your work will never be wasted!

LibraryLady · Post by **LibraryLady** » February 22nd, 2006, 2:14 pm

I like it Kara!

kayray · Post by **kayray** » February 22nd, 2006, 3:04 pm

I suppose it belongs on this page somewhere:

http://librivox.org/wiki/moin.cgi/TechSpecs

But, um, I don't really know where to put it so it's visible but doesn't make things look cluttery...

Gesine · Post by **Gesine** » February 23rd, 2006, 2:47 pm

OK. I still think it's too much to read, but we seem to have reached a general consensus and I'm happy with that. Now we need to let the world know. I propose the following:

1) We update all 'How to be a Book Coordinator' references and ask that this paragraph be included in all future projects. I think it needs to go into each intro thread because many (most) people won't read what's in the linked info (and I know this because I get many questions that are clearly answered in the recording notes to which every project links).

2) I was going to put up a couple of 'templates' intro thread (including code to copy) for BCs on the wiki, with all relevant info. We can update this as we go along. It can be amended as necessary, but it might be a good thing for newbies if the intro threads were roughly the same in layout and content, and in the way the claiming works (most BCs now use continuously numbered "sections" for projects with several parts with sub-parts or chapters, for instance). Of course the 'templates' can still be amended.
We can refer to these templates in the BC sticky references etc.

I'll have a go at number 2) in a minute and will post a link here, then everyone can bash it to death ;)