Ben Peoples Speaks


Audio Transcription
September 2, 2010, 9:38 am
Filed under: Uncategorized

Over the last weekend, I recorded a lot of audio.  I was looking for a way to easily make that unedited audio searachable, or at least indexable.

I came across several solutions:

So I took a chunk of audio, and tried transcribing it various ways.  The most accurate was, of course, manual transcription, followed by Mechanical turk.


Google Voice:
Hi Barry, I don't know, like. Randy time particularly, couple to delete them. Well know if guys bye yet but he it, it was. When I remember it in something right at the end of the is nothing. There's, bye. That Walk along. Yeah. Yeah it is really have a lot. Gary, I hate to even a lawful attic. He, god, all, hey this is not. Anyway, I love you. Love you guys would say I don't know what is that I sent you a little Yeah, Thank you. Have a great deal that you've evidently there up later. Right at the top exit poll the. Bye. Later, this Yeah, the birthdays popped into my head and I had no idea. I'll be getting in at Hi there, It's Tuesday and a just been better on that. The first when you can. 833. That's my bedroom for my birthday. Well, I was away. One day I Yeah, Mom and I really think that what we can get that. Dr meeting you. All of the publicity for the that bridge when I C voicemail at all SPCA. Pat your offer valet. Yeah, hi that have been like ohh i don't remember the first week here, hey he left to go pick you up. I don't like what do we need you to be at least an hour and a half HI. You see, if you get a van and I did, the other when it comes to mind. Again, I don't know. We love you. What flight you that you're from. California for my birthday, and I think the excitement door at my bedroom that your birthday present. For me it was it with you. That was all of us. I I I don't know if there is a hold of you Tuesday. The. But. Everybody have fun. Those were the to do that. I remember it being like the most unexpected. Cherie yet definitely hey. But yeah.

64k-word (CMU Sphinx, 64k-word dictionary):
our air and aol are hopeful that the league and will well and then i fantasize that ireland again but he knew it is the way they are in an area is the sole envy and news updates there is a sign that says walk along a busy day and i had a game fill a dandelion gary's a santa fe is a walk along venues for a shade ulcer did all it is not a loan deal and we didn't think he did that we are losing their fifth minute on an issue that hit the hole again this bank had a career as a means to an end and then a player who led the top of the profits of writing and in the last half its the crusade of how clinton and annan had delayed a vote the inventiveness iler is assisting said including disney deserted monday yet there have not the first is the new a three day period that made it through for my birthday well it wasn't in school and an iron hand analysts yet his hand and now nominated new dehere eighteen attendant who hinted that drive the new low all of her little didn't seize heartbeats dobrinja was icy and now it was in love and all effete and we are delighted that i have a late night and this piece for her eighty is that you're on ramallah came piano and let it be licked all and remembered the words the move against the time he left a note in knocking off apparently worked the ninth and he flew in the indian police in our half of the idaho i didn't get to eat them he didn't really need not done a guiana island that closed in london and again i don't know who they knew this home time lutes flying you next year from california for my birthday and a big thing and are in their bedrooms and the your birthday present for me as a sign that the few now a dollar last man in a sudden there's olivia de fazio

5k-word (CMU Sphinx, 5k word):
our air at eight article in the league in well right then a fan is on the island yet but the new it is the way they are in an area is some energy and news of its days assigned this is a walk along a busy day and i had a game fill a satellite gary k. i. n. a. as they walked along any food aid and details and did all it is not a loan deal and we didn't think he did that we are losing their fifth minute on an issue that hit the hole again this bank can be renewed for a minute and then a player who led the top of the house is going home with a left activists the first in a hospital and annan had no idea how they can then and this time it says it isn't included in a very creative long dead yet and not the first is the new a three day period that made it through for my birthday well it wasn't in school and an iron hand analysts yet his hand and known nomination eighteen day period in the standings going into a van driving you all all of her little day to a cease fire between the three engine was nice seeing that they know it was in love and off her feet and we are in a light and lived a mile and a few feet from her duty is back to her down and dan and minutes at the lake all in remember the words the move against the kings in the last thing up and knocked him off apparently worked the ninth and he flew in the indian police in our half of the island i didn't get to eat them he didn't really need not done a p. l. o. in a closed in london and again i don't know who they knew the club i lose it's flying you've been a year from california for my birthday and having been signed and or in a bid sunday your birthday present for me as a sign that the fuel now a dollar last man in a sudden there is no via these days he added

Manual transcription:
Disney world, or um I dunno, just like random times
Particularly coupled with the silly dad walk
Well, right that's what I was talking about Yeah, but he it was when i remember it was somewhere at mgm there's a sign that says wakka wakka this way and he would do the silly dad walk every time he saw that sign, and would say wakka wakka this way
We'd all start doing it too Mom would be walking along behind us you guys are weirdos
Yes, I am not with them.
Just like Sam Eagle That can't be the craziest thing he's ever done
But that's up there
But one of the top things that pops into my head when you ask that.
The first thing that popped into my head, and I have no idea how big a hand dad had in this well, actually, it was two things and it could have been dad, or it could have been mom and dad. together uh, the first one was when you guys redecorated my bedroom for my birthday while I was away at school one day. that was rediculous
That was um mom and I did the redecorating but dada was the the one who did the driving you all over the place The detour because the bridge was icy even though it wasn't. at all becuase we needed like an extra 45 minutes. he was driving you home from Ballet
Yeah, I remember that trip being like "wow, I don't remember detours taking this long"
Because mom and I were like "we're not done", we need you to be at least an hour and a half.
"I can do that"
And so he did! He did a really good job
The other one that comes to mind And again i don't know whose idea this was uh was flying you back here from california, for my birthday and having a big sign on the door to my bedroom that said that your birthday present for me was inside, and it was you
That was all of us
I had a suspicion that it was all of you
I don't remember whose idea it was, but everybody had a part in that.
those were the two that I remember as being like the most unexpected in terms of crazy.

I was able to transcribe it at around 1/4 full speed (that is, 4 minutes per minute of audio).  Mechanical turk was a slower transcription, of course, but didn’t involve me.  I chopped the audio into short segments and created HITs from those.  The rough cost to get it done relatively quickly was $0.01/second of audio, and was able to transcribe 55 seconds per hour at what I’d call 99% accuracy.   I think at a slower rate (maybe 15-30 seconds per hour) you could reduce the rate.  Transcription at 1 cent per second was very popular, particularly with the short 10-second segments.

The next experiment will be to upload 60-second segments and ask for keywords or summaries.

Advertisements

Leave a Comment so far
Leave a comment



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s



%d bloggers like this: