Re: Anyone like to transcripti WOWODC podcasts via Mechanical Turk?
Re: Anyone like to transcripti WOWODC podcasts via Mechanical Turk?
- Subject: Re: Anyone like to transcripti WOWODC podcasts via Mechanical Turk?
- From: Ted Archibald <email@hidden>
- Date: Fri, 13 Jul 2012 04:55:04 -0600
You can use MTurk by using
crowdflower.com outside the US. I'm in Canada and I use WO app that submits jobs and pulls data from MTurk via crowdflower. I highly suggest it. It makes MTurk far easier to use as well.
That being said, wouldn't it be easier just to shoot the files to a transcription company? Especially since there's alot of domain specific terminology.
The biggest pain I've had with crowd-sourcing is getting your error rate down. You can brute force it with high cost jobs or job repetition (which crowdflower does beautiful). Or you need to manually review everything. Or you can do it with trial and error, and how your job is designed. I only got the hang of it after submitting 6-12k jobs per month for at least 4 months before I was really happy.
On Fri, Jul 13, 2012 at 2:56 AM, Amy Worrall
<email@hidden> wrote:
On Fri, Jul 13, 2012 at 9:11 AM, Johan Henselmans <
email@hidden> wrote:
> Some people reported having transcripts done this way. Unfortunately, I am not allowed to use this tool, as I am not a citizen of the US of A.
where they submit the audio to Mechanical Turk and collate the
results. It wasn't too expensive and it was a lot better than having
to transcribe it myself!
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (
email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to
email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden