legamus.eu

One way is to invite other users (I've found that, hurray!). But I think it's better to have one log in for all Admins/MCs. We need an email address for that and I think it should be one connected with Legamus. I've asked Viktor about that.
One way or another all Admins/MCs will have access to the blog/catalog. Naturally!

Viktor wrote:Strange, because mp3gain homepage describes how to change the volume. That does not work?

No, it does not. I reinstalled the program to make sure, but the volume remains the same as far as I can tell.

I have not yet had time to look at the tags programs, but I've had to upgrade my OS, meaning I now have access to the App Store, there might be something there as well.

I've made an account at WP, username: neckertb (surprise).

"Replay Gain" does not really change the volume, it just adds a hint in the MP3's metadata. If the player does not honor this, you won't notice any difference. Apple stuff (iTunes / iPod) uses a different metadata item for the same purpose...

VLC should honor the metadata. Do you hear a difference between your original and your modified file with VLC? Maybe the difference is also very small, and we won't hear it.

To me, it seems that audio normalisation, which touches the real data and will be honored by any player, is more important than this fiddling with metadata. Do we already do that as a standard step in the recording/postprocessing?

Ideally, the files submitted by the readers should be 89 db. So yes, they should be normalised.
However, we know by experience that:
1. telling the volume in db can be difficult
2. not all readers bother or don't know how to do it.
So at LV, we MCs often have to homogenise the volume. also, for group recordings, the volume can vary quite a bit from a reader to the next.
It might not be a big problem here since so far it is only experienced LVers that have signed up, but it is sure to come up at some point.

So in the long run, it would be best to have an uploader that

• adds metadata from the database to the ID3 tags
• re-encodes in 64 kbps
• performs audio normalisation "if needed" (if the data is "too far away" from a normalised version)
• sends the result back to the MC or, if possible, directly uploads to a freehoster

Right?

Exactly

Normalizing means amplifying to a certain volume?
This can be dangerous, especially when the original value is far away from optimum. With very low files, background noise gets amplified, too, and it needs a human listener to determine if noisecleaning is needed.
Many files can't be amplified without clipping, which distorts the voice and can damage the phones.
A tool that checks the volume would be great, but actual change without a human ear, I don't know.
We should better have a narrow range of OK volume (like 87 - 91 dB) and help our volunteers (readers, BCs and PLs) to check the volume and make adjustments to their recording set up and editing process.

mp3gain is not the very best, but at least it doesn't change the file.

Everything else on the list would be just wonderful!

The database should contain some more information than the LV Magic Window.
Track number; section title; book/project title; author; text source
book/project title; author; text source will be the same in all section for most projects and can be mass set. But for collections it would be great if we could enter them for each section.
Depending on how the catalog search works, separate author will be necessary anyway.

The metadata tool should add genre speech and pd info for the text and license info of the recording in the comments.

re-encodes in 64 kbps

I'd make sure to test this thoroughly for quality issues. We found when submitting samples to publishers at 64 kbps that the Audacity LAME encoding from a 44 KHz recording to 64 kbps MP3 produced poor audio quality. I got around it by recording the samples at 22 KHz sample rate - then the 64 kbps MP3s were of much higher quality.

I only flag this up - I really don't know what I am talking about and my thought processes are a bit muddled at the moment for various reasons.

I don't know what encoder the Internet Archive uses, but I think it has been improved recently.

Ruth

neckertb wrote:1. telling the volume in db can be difficult

I actually don't know how to tell the absolute volume from Audacity relative dB-level. Could anyone tell me that? The loudest parts in my recordings are (after compression, no normalisation) at appr. 0.6 in Audacity; that seems to be OK for Librivox. But would be nice to have a more precise way (also if I am to judge other people's recordings in the right way).

LV uses mp3gain http://mp3gain.sourceforge.net/ in the cataloging process, but it doesn't work on Mac.
Phil has made a tutorial video for it http://www.youtube.com/watch?v=toef8_-qsms

Celebration! Our first book is finished and cataloged!
http://legamus.eu/blog/?p=48

Of course it didn't work the way I thought it would. The plan was to create as separate page for every book. But with that we would have no tag cloud and no categories. So I made the entry a post (instead of a page). The HTML is the same as in the drafts.

The book is only 1 file, so I think we can offer 128 kbps for download. No need to make it smaller.
Note: I made some small changes to the ID3tags, so the link is not the same as Piotrek's upload link in the project thread.

This is the first real thing besides the drafts and tests. Much room for improvement, I'm sure.