Suggestion: Add Producer Role

Comments about LibriVox? Suggestions to improve things? News?
JordanN
Posts: 24
Joined: April 13th, 2014, 2:18 pm

Post by JordanN » April 19th, 2014, 5:09 pm

The producer would accept each contribution and master it, turning it into a nice neat package before handing it over to the coordinator to be added to the main library, just as a producer would on a commercial project.

Producer responsibilities:
  • *Mastering. Applying compression, EQ, and other effects so that audio is sufficiently crisp, loud, and understandable. Make the audio as "professional" as possible.
    *Consistent filename scheme.
    *Adding ID3 tags to each file so that each is identical.
    *Ensuring there are no inconsistent effects or distracting elements.
I suspect that there are plenty of folks out there with audio editing experience who might not be interesting in lending their voices, but would volunteer for this role. This would also free up the contributor to focus on delivering quality virgin recordings, without being distracted by whether some non-standard obscure third party tool thinks their recording is the perfect volume.

TriciaG
LibriVox Admin Team
Posts: 47793
Joined: June 15th, 2008, 10:30 pm
Location: Toronto, ON (but Minnesotan to age 32)

Post by TriciaG » April 19th, 2014, 5:33 pm

*Consistent filename scheme.
*Adding ID3 tags to each file so that each is identical.
For the record, these two items are handled automatically at cataloging time.
without being distracted by whether some non-standard obscure third party tool thinks their recording is the perfect volume.
I'm not sure how this would work. Would this producer have to listen to each recording to determine its "to my ears" volume, then adjust the files accordingly? In that case, it could only be done after all the files are in, since the "to my ears" volume couldn't be pinpointed between different production sessions unless third party software is used. (And we DO use third party software to determine relative volumes. At cataloging time, we adjust the volumes to be consistent, if there is no clipping involved on quieter files. The system we use, however, may not work in all audio players, but it's the best automated system we've seen.)
Mystery/PulpFic: Dope, by Sax Rohmer
The one that started them all: Self-Help, by Samuel Smiles
Elizabethan Poetry: The Psalmes of David
Boring works 30-70 minutes long: Insomnia Collection 5
Short essays: Elia, and The Last Essays of Elia

TriciaG
LibriVox Admin Team
Posts: 47793
Joined: June 15th, 2008, 10:30 pm
Location: Toronto, ON (but Minnesotan to age 32)

Post by TriciaG » April 19th, 2014, 5:36 pm

Disclaimer: I'm of the "shoot holes in the idea and work out its issues before saying it's a good idea" personality. So just because I'm picking at it doesn't mean I'm rejecting the idea outright.
Mystery/PulpFic: Dope, by Sax Rohmer
The one that started them all: Self-Help, by Samuel Smiles
Elizabethan Poetry: The Psalmes of David
Boring works 30-70 minutes long: Insomnia Collection 5
Short essays: Elia, and The Last Essays of Elia

DrewJ
Posts: 830
Joined: July 16th, 2013, 5:30 pm
Location: Memphis
Contact:

Post by DrewJ » April 19th, 2014, 5:55 pm

So is the point of this role to give people who aren't able to record another way to lend a hand?
Or is the point of this for less technically able readers to be able to submit a recording and not have to worry about the finer points of compression and noise removal?

As my dad used to say, "If you can't measure it, you can't manage it." And I feel this applies to sound as well. Like what Tricia said, without the use of a program that measures decibels, there is no way to have all recordings conform to a standard. For instance, I enjoy things loud, not so loud to harm my hearing, but loud enough to really blast from whatever I'm listening through. A friend of mine (who isn't on Librivox... yet) hates really loud audio, and prefers audio to be soft, and then amplify it in her pc/mp3 player. Even here, 'loud' can't be measured.

The standard of Librivox is 87 to 91 I believe. So anything over 91 decibels is 'loud' here at Librivox, and anything under 87 is 'soft'.
If I was in the role of a Producer, I would (to my own tastes) probably keep the volume of recordings around 95.
Do you recommend a different program to check the average volume of a track against the official Librivox volume range? Personally I find Checker very useful, since it not only catches volume issues, but also format, clipping, etc.
When the hurlyburly's done,
When the battle's lost and won. -Second Witch
Read some poetry?

JordanN
Posts: 24
Joined: April 13th, 2014, 2:18 pm

Post by JordanN » April 19th, 2014, 6:35 pm

I admit that I'm mostly annoyed with the 89dB specification, which seems to refer to dB-SPL (volume relative to quiet), which isn't relevant as we have no idea how loud the listener's speakers are or what their noise floor is. We can't control the listener's environment.

What I'm suggesting is that we measure volume relative to maximum, which is how all of the DAWs and other professional audio equipment (so far as I've seen) measures it. This is why when you look at a sound board, unity (0) is at the top, and most of the numbers are negative.

What I'm suggesting is that we adopt actual industry standards, such as -23 LUFS (or EBU R128, referenced in http://en.wikipedia.org/wiki/Loudness_monitoring) which is an actual loudness standard. These should be easily checked in all audio editing programs.

If you Google -23 LUFS, you'll find a lot of articles to explain it, but here's one such result: http://www.tcelectronic.com/loudness/loudness-explained/

Article about reading meters on DAWs:
http://therecordingrevolution.com/2013/11/25/do-you-know-how-to-read-your-meters/

A Sound on Sound article on loudness monitoring:
http://www.soundonsound.com/sos/sep11/articles/loudness.htm

Another article:
http://designingsound.org/2013/02/loudness-and-metering-part-2/

Folks have submitted feature requests for the Audacity VU meters to comply with EBU R128 (http://wiki.audacityteam.org/wiki/VuMeter).

What I was getting at with the produce role suggestion is that it's hard to get every contributor to record identically. We should focus on contributors keeping a consistent distance from the microphone and eliminating background noise.

It might be better for some folks to specialize in handling the mastering stage. No, you can't produce miraculous results in post, garbage in, garbage out. However, with quality material coming in, the mastering engineer/producer would be able to polish it into an outstanding "product".
Last edited by JordanN on April 19th, 2014, 7:06 pm, edited 2 times in total.

ToddHW
LibriVox Admin Team
Posts: 11950
Joined: August 14th, 2011, 4:24 am
Contact:

Post by ToddHW » April 19th, 2014, 6:57 pm

We already do these things; you just have not been here long enough yet to see them in action across the full range of forum threads. Because we are attempting to record everything in the PD as our vision, we have distributed the responsibilities across all our readers instead of focusing them in a few overworked people. This also means that people hear what THEY submit going out to the world rather than some processed variant - a source of pride I think to every one of us.
*Consistent filename scheme.
*Adding ID3 tags to each file so that each is identical.
As TriciaG pointed out, these are now handled automatically at cataloging. We still ask for them cuz we haven't updated all our instructions yet. And I know I still like having them in my files before they get uploaded so I can better track what it is that I have on my computer drives.
*Mastering. Applying compression, EQ, and other effects so that audio is sufficiently crisp, loud, and understandable. Make the audio as "professional" as possible.
*Ensuring there are no inconsistent effects or distracting elements.
Our PL process handles this, with PL folks commenting on loudness (using tools such as the replaygain add-on for Audacity), understandable, and distracting elements so that the individual readers can fix these issues themselves. While our standard PL definition of "acceptable" is low, I think most folks want to exceed it and ask for more help from those with some experience than we would have imposed upon them externally. Often you will see this in forum threads as coaching of new readers on how to adjust volume, use compression, clean out noises of various sorts from discrete elements that need notch filtering to hiss that Audacity handles very well. We have some great videos that enable folks to learn how to improve their output. The more readers we teach to make "acceptable recordings" the more we can get into our catalog.

The "Make the audio as "professional" as possible" that you request we do not do. That is what pro's can afford to do but that we amateurs will only do the extent that we each enjoy learning new skills, and to the level that we are each comfortable with as we fit this hobby into the rest of our lives. We usually stop at the "sufficiently crisp, loud, and understandable" standard that you note - with a strong recognition that "sufficiently" can be very hard to define.

Thanks, Todd

(Note - you might want to take a look in our Dramatic Works forum. The role of "editor" that is required there to combine the various reader inputs together into a unified product might involve some of the attitudes and skills you are suggesting.)

JordanN
Posts: 24
Joined: April 13th, 2014, 2:18 pm

Post by JordanN » April 19th, 2014, 7:15 pm

Thanks for the reply, Todd. I'm sorry about the impetuous new guy syndrome.

I admit that I'm mostly annoyed about the ReplayGain local standard (ReplayGain is a program designed around a single man's opinion), and strongly propose that we adopt the EBU R128 international standard instead.

philchenevert
LibriVox Admin Team
Posts: 19447
Joined: October 17th, 2010, 9:23 pm
Location: Basking by the Bayou
Contact:

Post by philchenevert » April 19th, 2014, 7:21 pm

This is an interesting thought Jordan. i am referring to your proposal of adding a 'producer' to edit raw files for those so inclined.

My first thought is that this would indeed be a boon to people who just want to record and not learn all that technical editing stuff and also to those people who might wish to contribute to LibriVox by being producers (for want of a better term). I assume this person would also serve as a proof listener for the files.

My first thought is that a good deal of back and forth would be necessary between producer and reader for each section before it is read to be cataloged. Every section I've every done (and I've done more than 1300 of 'em) has needed words, phrases or sentences replaced during editing because I messed it up somehow. And some have needed many replacements. True, this may be a product of my method of recording but I'll bet that most everyone has to replace words and phrases during editing. Anyway, this would mean the producer would need to do their magic, then wait for a replacement word or phrase or file then do the whole thing again. Is this how you would see the process? or would you see the reader doing a first edit, replacing words, etc, then uploading it?

Thinking about the process or recording, I think that actually listening to a track that I have recorded while following the text is extremely good for me as a reader both to hear the rhythm and 'know' when something isn't right. Reading is a skill that improves over time but only if we hear ourselves and can see how to improve. Removing this would not be a good idea, but having someone to check for the technical issues you mention would indeed be nice.

Sound: Yeah, we use what I call a 'perceived loudness' scale of some kind that is used in MP3Gain and our Checker program. It does not have any relation to industry standards that I know of but the darn thing works somehow. Moving to LUFS or some other standard would not be difficult for me since I just use my Audacity wave form pattern and can get it pretty close, but would throw our instructions into a really big ball of tangled thread.
ImageImage
video of our Zoom 10/18/2020 https://youtu.be/OtvgxviSrTw

My horses' name is Mayo. Sometimes Mayo neighs

ToddHW
LibriVox Admin Team
Posts: 11950
Joined: August 14th, 2011, 4:24 am
Contact:

Post by ToddHW » April 19th, 2014, 7:38 pm

JordanN wrote:Thanks for the reply, Todd. I'm sorry about the impetuous new guy syndrome.
I was not implying any disrespect for new guy impetuosity. I just wanted to point out that we DO do what you suggested but it is not obvious unless you have seen, for example, Phil spend weeks of coaching trying to help a new reader learn to handle noise issues in their submittals on a specific project. This is not something you see in the easily reviewed forums like Help Wanted, but is buried in the individual project forums where most readers spend their time.

(Though probably resulted in another of his fine videos....)

Thanks, Todd

philchenevert
LibriVox Admin Team
Posts: 19447
Joined: October 17th, 2010, 9:23 pm
Location: Basking by the Bayou
Contact:

Post by philchenevert » April 19th, 2014, 7:43 pm

Thanks for the support Todd. Yeah, we do a lot of coaching and teaching here. This thread has urged me to post about Google Helpouts. (see another post somewhrere) We might be able to offer them as part of our tutoring. I find them very helpful and easy to use (once i got the camera to work). There is nothing like actually talking with some struggling to understand compression, see their file and show them in real time how to do it.
video of our Zoom 10/18/2020 https://youtu.be/OtvgxviSrTw

My horses' name is Mayo. Sometimes Mayo neighs

MaryAnnSpiegel
LibriVox Admin Team
Posts: 15992
Joined: February 23rd, 2009, 4:37 pm
Location: Chicago, IL

Post by MaryAnnSpiegel » April 19th, 2014, 8:04 pm

Jordan,

After I edit my files, I don't have patience for anything except noise reduction and amplifying. As a result, I send most of my files through Auphonics with a loudness target of -20 LUFS. Sort of "producer in a can" for me.

I check a fair number of the test threads and it seems that new readers either have it all pretty much down from the get go, or they are struggling with the basics. I'm happy for those with no experience in this sort of thing to get a file with the correct technical specs, sort of close on volume and moderately low background noise. Expecting more when they are just learning how to read and edit their files (remove and fix mistakes in words read) would be a stretch. It's a pretty steep learning curve for those who haven't done it before. For those who have moved up the learning curve, it might be nice to get some coaching from a producer on more things to do with their files to make them sound better. I turned to auponics because I have no idea how to think about compression and the like.

I don't see us switching to a LUFS target because our cataloging software is set up to use MP3Gain for a dB target, but I would appreciate it if you would consider writing some articles for our Wiki about how a producer would approach an audiobook project - preferably using Audacity because that's the most common recording software around here. Having a good resource for compression, leveling, notch filters, high pass, low pass, and all sorts of other things that are mysteries to many of us would be a benefit to the whole community.

MaryAnn

RuthieG
Posts: 22011
Joined: April 17th, 2008, 8:41 am
Location: Kent, England
Contact:

Post by RuthieG » April 20th, 2014, 12:20 am

I am all in favour of improving recordings, but bear in mind that at any one time we have 500 - 600 projects in progress, some of which take several years to complete. Were you suggesting that there should be a producer for each of them?

Ruth
My LV catalogue page | RuthieG's CataBlog of recordings | Tweet: @RuthGolding

ekzemplaro
Posts: 2030
Joined: December 31st, 2011, 7:17 am
Location: Tochigi,Japan
Contact:

Post by ekzemplaro » April 20th, 2014, 4:53 am

Hello Jordan san,

Welcome to LibriVox. I hope you enjoy it here.
JordanN wrote:I admit that I'm mostly annoyed about the ReplayGain local standard (ReplayGain is a program designed around a single man's opinion), and strongly propose that we adopt the EBU R128 international standard instead.
A good suggestion.
So I'd like to know the difference when the poposed standard is applied.
1) If there's a program to evaluate the recording using the proposed standard, please let me know.
I prefer a program equivalent to mp3gain.
Then I'll check my 100 recordings and how many recordings become NG by the proposed standard.

2) If there isn't a program yet, can you check it?
Pick up 10 LibriVox recordings (you can use my recordings), and let us know how many recordings become NG by
the proposed standard.

I bet the NG ratio is within 10 percent.

Cheers,
Masa

dlolso21
LibriVox Admin Team
Posts: 4556
Joined: January 11th, 2011, 12:13 pm

Post by dlolso21 » April 20th, 2014, 6:00 am

Jordan,

Welcome to Librivox!

There is always help needed in the Dramatic Works area for those with audio editing skills. Your assistance there as an editor would be greatly appreciated by those who like to record, but not to edit/produce.

We are a very diverse and widely dispersed community with a wide range of talents and equipment so setting one standard that all can meet can be difficult.

I like the idea of a loudness meter in Audacity since many Libirvox volunteers use that program and will most definitely test it out when the feature is implemented.

David O.

JordanN
Posts: 24
Joined: April 13th, 2014, 2:18 pm

Post by JordanN » September 15th, 2020, 7:05 pm

It’s interesting that I’m revisiting this discussion several years later, but I yet still mostly agree with myself.

[*]Executive Producer role: a person committed to the mission of librivox. She should have good ears and a business sense and a solid relationship to our user base. Probably would be a double roll with Admins.

[*]Project producers (Moderators) receives direction and suggestions for how to organize a project to be most effective. They give final feedback to the mastering engineer before a project is published. The coordinate with volunteer talent (readers) to gather the scripts recorded.

[*]Mastering engineer - in music production, the mastering engineer receives a package of mixed audio files from the recording engineer. She then uses high-grade equipment and detailed tweaks to make all of the albums flow together into a cohesive album that can be enjoyed in the car while driving, relaxing at home, played to a classroom with professional audio equipment, or played from an iPhone into Airpods. We don’t know the circumstances of audience, but we can put an effort forth to maximize the quality of the projects that we are releasing. Part of this will be accomplished through the leadership of talented mastering engineers, and the rest through the standards that they agree on. The Librivox Lead Mastering Engineer would write the production standards for Librivox projects; to ensure the best quality according to industry best practices, give the readers clear guidance to help record awesomeness, and guide the other Editors/Masters.

[*]Editor - mentors readers without much technical background to help them gain the best quality from their available tools. Does light editing to maximize the quality of that recording or requests a replacement recording if it’s unusable.

[*]Masterer - Edits all of the individual readings together into a cohesive project, ready to be distributed as a podcast or other such formats that our users need.

Loudness standards. The industry has migrated to LUFS as a standard way to characterize loudness. All of the DAWS should have support for LUFS metering now. Though librivox is a volunteer organization, our listeners also listen to commercial and your volunteer editors likely do projects outside of librivox where industry norms are expected.

(From: https://www.edmprod.com/lufs/)
Service Loudness Target
Spotify -14 LUFS
Soundcloud No loudness normalization
Apple Music -16 LUFS
YouTube -14 LUFS
Amazon -11 LUFS
Beatport/DJ Stores Recommended -6 to -9 LUFS

Post Reply