HP:MoR Audio Book Pilot

by Eneasz1 min read29th Mar 201130 comments


HPMOR (discussion & meta)
Personal Blog

Like far too many people, I have a job that for decent chunks of my day doesn't require much concentration. I listen to various podcasts and ebooks to alleviate the boredom at these times, as well as to get me through my daily commute. I have several friends with plenty of audio-listening time, a fair bit of which is given to amateur podcasts (in the "not getting paid for it" sense), who have much less time available for straight reading. I have, on more than one occasion, thought "Wouldn't it be great if Methods of Rationality was available in audio format?"

Well, now it is.

At least, the first chapter is, in a "testing the waters" sort of way.

If you have 17 minutes free and any interest, here's the file - http://www.filedropper.com/hpmorpilot

I have one over-riding question to anyone/everyone: is this of any value? Would this be useful to anyone here, either personally, or to share with others? Would a spruced-up first chapter, and regular production of further chapters, be something you want more of?

A closely-related follow-up question: if so - am I the right person to do this? Even "acceptable", I'm not shooting for perfection. Or is something about my voice/manner so grating I should stick to non-radio work.

If the answer to these is "yes", then are there any comments/suggestions? I'm new to this recording thing. I got a decent mic and a pop-filter, and I've played around with the software a bit, but I'm still learning. I have noticed that I need to enunciate more, it turns out my "can't win" sounds just like my "can win". Further suggestions? I'm thinking of varying my pitch when different characters are talking to make it more apparent when dialog is switching back and forth, hopefully that won't be too cheesy.

If people want more of this I'll get an actual server to host the files and make it available through iTunes as a podcast.

Again: http://www.filedropper.com/hpmorpilot


30 comments, sorted by Highlighting new comments since Today at 6:39 AM
New Comment

You may or may not be aware of this, but there is a decently large community of people who record audiobooks for free: LibriVox.org. Their forums have a lot of resources and technical advice on how to do this, and they're quite good at offering technical feedback as well.

I would strongly advise you to submit your work there to receive more expert comments, and quite possibly recruit some new reader (meaning both "expanding MoR's audience" and "getting interested volunteers for the audiobook").

I'd actually be interested in participating, but I understand perfectly what a hassle it would be to piece together multiple voice actors. Dibs on Quirrell if you change your mind, though!

Agree on the usefulness of having stuff in audio format.

Your voice is easy on the ears, so that's good, and of course you'll learn to enunciate at all times as you pick up more practice. Try a greater use of emphasis on key words or contrasting words - that is, speak in italics. (I write with italics way too much partly, I think, because I have learned to speak with italics when recording for podcasts.)

What's the background noise? You might be able to remove some of that even with something simple like Audacity. Ask if you need details.

Doing multiple for voices for multiple characters is for pros. I've never been able to do it competently, so I have no advice on that.

Audacity is what I (try to) use. Details would be helpful! I meant to play around with it more last night but was side-tracked. What sort of background noise were you picking up? The intermittent stuff is probably my inhaling.

During each recording, I record 30 seconds of silence. Then when editing, I select those silent 30 seconds and choose Effects -> Noise Removal and choose Get Noise Profile. Then, back on the main editing window, Select the entire track (Ctrl + A) and choose Effects -> Noise Removal -> OK.

Works pretty well.

A friend and I thought of this a few months ago, and we actually recorded the first chapter. Then we calculated how much time the rest would take and concluded that we simply weren't willing to make the investment.

That said, would you be open to collaboration, to defray the huge time requirements?

I think so, but I'm not sure how that would be done without taking up even more time. Unless you meant alternating chapters?

Sure, alternating, or spreading the load among anybody who has a microphone and wants to contribute.

This would introduce issues of quality control - you might have to tell someone that their reading was just not very good.

The consistency between tones and character voices might also be jarring.

On the other hand, a little variety might be a fun exercise. Has there ever been a "community audiobook?" (Cursory googling suggests not.) We could be the first. Anyway, I'm sure there are plenty of people who couldn't stomach twenty-plus hours of reading into a microphone, but would be happy to do twenty minutes, or one hour.

I'd be willing to do a specific character's voice or two. As long as someone edits the audio there's no reason it has to be recorded sequentially. (If anyone wants to know what I sound like, I did the voice of Meri Lin (first voice heard) on this Youtube commercial for a friend's webcomic; it's not a good-quality recording, and the timing was awkward to pace with the video so I think I sound stilted, but there it is.)

Thank you! :) You have a nice voice, and I would love to have a woman for the female parts. Unfortunately, for the reasons I laid out in my reply to moridinamael, I'm going to decline for now. If things change I'll send you a pm to ask if you're still willing. Thanks!

For a while I had a dream of having a different person for every major character, but for an amateur project this presents many problems. The first being geography, it's unlikely we can all meet in one place for this. The second being scheduling, it's hard for many people with other full-time obligations to meet regularly and consistently without "it's my job" working as a good excuse. The common response/solution is to have different parts recorded separately and sent in to be recombined, but that leads to two new problems.

First, tone. Without people there in the same room to play off each other the different readings can be jarringly out of character with each other. Secondly, it's vastly more post-production to recombine everything (which I'll get into next).

Alternating chapters would be great, but I think I'm going to decline for now, partially for the reasons you mentioned. Having a consistent tone is probably a good thing, and I would have a very hard time telling someone I couldn't use their reading. For someone to spend $100 (or more?) on recording equipment and several hours reading and be then told it's no good seems... cruel. But also for two other reasons (I'm all about two's today)

  1. Post-production. This is the majority of the work. It took me about 2 hours to put together the 17 minutes of the pilot. I expect this to go down drastically as I get better with the tools, but post-production will always be more than half the total time spent because I have to go back and re-listen to the audio in it's entirety while pausing repeatedly and making tweaks/changes/removing flubbed lines/inserting clips.

  2. Reading is the fun part. I really enjoy the actual reading into the mic part of this. :) So while it's time consuming, it's something I would consider to be fun in it's own right. If someone else were to send in audio of them reading then I still get all the work of post-production without the fun of reading aloud to compensate for it.

Hopefully this isn't burning any bridges, and in the future if I'm able to work things differently or realize I really need help I'll still be able to come back and ask for some. I do appreciate the offer.

Geography isn't a problem. Skype can be used to get everyone "together."

No bridges burned - it's your project, and if me and my friends had intended on doing it, well, we would have done it. I agree with all of your objections. If you want this done "professionally" then you had better do it yourself.

If you change your mind and decide you want help, please do ask!

I really like this. You're well above "acceptable", there are a lot of bad audio book readers out there! I agree that differentiating the voices would be valuable, and obviously editing and work on enunciation would help but this is a great first try. Please make more!

more feedback: I think the echo chamber is somewhat over the top, and the shout at the end really got me because it was actually a shout and I wasn't expecting the audiobook to get significantly louder than it was.

I might suggest using a different voice for "Narrator" who can describe things that happen and harry's thoughts, to differentiate from his dialog without going all echoey.

I have one over-riding question to anyone/everyone: is this of any value?

I think so. Audiobooks are great for long car trips. As for your other questions - your voice is great, but you need some practice. And some editing - there were a few noticeable stumbles that ought to have been corrected in a professional sounding product. If I had to offer one piece of advice on your manner, I would probably suggest that you try to make it more 'theatrical' - don't worry about being too 'cheezy' - cheese works in this format.

I was a little surprised that you abridged and modified the text so much. Is that standard in audio books? And there was one thing I didn't like - the echo-chamber effect for Harry's thinking to himself. I suppose you need some kind of effect, but I didn't care for the one you chose. But that is a minor point. Well done!

I was a little surprised that you abridged and modified the text so much. Is that standard in audio books?

Yes, it is common... but it shouldn't be! I can't stand abridged versions.

Actually the "abridged" part is in reference to the flash-forwards that start out several of the early chapters (I just don't know how to do those in a way that doesn't sound confusing). The rest was read straight, directly from the most recent version available at fanfic.net. Eliezer changed the original text of the first chapter a fair bit, which also surprised me when I went back and re-read it in preparation for this.

I'd definitely fix the stumbles in any final version I put out. Is the echo-chamber OK for the letter-writing portion?

Thanks for the feedback! :)

Eliezer changed the original text of the first chapter a fair bit...

Ah, you are right. I was looking at an old .pdf copy when I accused you of excessive abridging. As for abridging the flash-forwards, you probably made the right choice. Though you could present them before the announcement of chapter name and title and other boilerplate.

As for the echo chamber, I would save it for things done in 'Magical Voice' - spell invocations, oracular pronouncements, sorting hat, etc. I do agree that the letter-writing and thoughts-to-self ought to be distinguished in some way, and perhaps in the same way, but the echo/reverb just doesn't seem right for them. They should somehow be less portentious, rather than more.

I would definitely say it has value, and I think you have the potential to make this really enjoyable. Unless you have an insane amount of free time, though, I think it would be a good idea for this to be more of a shared effort.

I have at least one friend who would read HPMOR if it were in audio form.

I have one over-riding question to anyone/everyone: is this of any value? Would this be useful to anyone here, either personally, or to share with others?

Sure, but no for me personally. As a matter of course I put my text based books through TextAloud. The quality is such that I've actually got fed up with 'official' audio versions of several books and replaced them with the computer generated version. In fact, given the characters in MoR the computer generated voice is probably going to sound more realistic! :P

Could you post a sample? I run Linux and tried espeak and festival and wasn't particularly impressed. Do you have any tricks you might share? I'm interested in this, but noted that computer synthesis can't interpret article structure and certain pauses, etc.

I wondered if you do anything special or just have TextAloud read a file as-is.

This! I had already emailed Eliezer about this, so I was delighted when he mentioned it in the preamble for the current chapter (71).

An advantage that hasn't been mentioned thus far: audiobooks are very useful for people who are visually impaired.

http://openetherpad.org/YvnjtzmvYi-EF-BB-BF Etherpad for organizing audiobook recordings.

I was just about to post a separate discussion query about something similar -- would there be any value to audio forms of the sequences? I prefer reading... but the sheer volume of material to digest and the convenience of using "downtime" (particularly transit time) is quite appealing to me.

Would anyone like to comment on this? Or is it different enough to be it's own discussion? I had similar questions myself re. whether I'm the one to do it ( I don't know that my voice is particularly appealing over long periods of time!) and if such a thing would be of value.

You may want to make a new discussion topic for this question, this thread is old enough I doubt it gets many views any more. But IMHO, yes, more media options is always more better. :)

Thanks for the reply. I'm a bit unsure about this anymore. After wedrifrid's comment, I gave a few open source text-to-speech programs a try, but didn't really like the output. I tried it on What do we mean by Rationality and noticed right away that it's hard to get, when listening, the "structure" of something like those first indented definitions.

I guess I'm now responding based on a completely different approach. Perhaps a human reader could add in slight prefaces to various sections, like, "We mean: Definition 1: Epistemic rationality... and also, Definition 2: Instrumental rationality..."

This might help. Still not sure if I think this project/suggestion would be feasible and if it'd actually be that helpful. For something more "intense" like going through the sequences, I wonder if audio format would allow for the same rate of comprehension as reading text. I'd love to use it when driving... but wonder how much I really retain when driving.

As someone that often listens to audiobooks while driving, I don't find problems with comprehension, unless there is something that is taking up a large portion of my attention. As long as I can basically drive on instinct and muscle memory, I remember it as well as if reading it. If there is something that I have to listen to, or read, then I generally either stop the book, or go back.

One strange effect for me though is that if within the next week or so, I hear a part of the book, I can tell you exactly where I was at that time, though I can't generally go the other way, and think of where I was in a book when I was last at a location. The brain is pretty weird/cool.

Hmmm. Interesting to know. I did listen to a lot of religious debates, but I'm not sure how to tell if I have great recollection. I could summarize the gist and certainly remember the overall points/strategies, though.

Debates probably aren't the best for trying to probe post-listening recall, though. They're all over the place and I've listened to so many on the same topic that they blur together.

Maybe I should just try this and see what happens.

Re. the knowing your location, that's really cool. I can do something like that visually -- if a quote struck me, I can remember quite long after where it is on the page and just riffle through keeping my eyes fixed on that part of each page and find it.