[RNLD] Transcriber issue resolved (sort of)

Mark W. Post markwpost at GMAIL.COM
Tue Aug 6 12:17:17 UTC 2013


Dear Listmembers,

Tom Honeyman found the 2-part answer to the problem I mentioned below 
immediately. (1) Transcriber 1.5.1 can't handle 24-bit files (2) 
Audacity was down-sampling to 16-bit.

I tried installing ffmpeg support in Audacity and exporting as a 24-bit 
.wav and Transcriber couldn't handle this file either, so it's pretty 
clear that this was the cause.

So, as near as I can tell the only solution is to make a 16-bit file for 
use with Transcriber, and then link the resulting .xml file to the 
original 24-bit file for further use. Tom mentioned that a small amount 
of shortening or lengthening might happen in the down-sampling process, 
but at least with a 2-minute file the lengths of the original file and 
the down-sampled file are exactly the same.

Cheers
Mark


On 06/08/2013, at 9:36 PM, "Mark W. Post" <markwpost at gmail.com> wrote:

Dear Listmembers,

I've been having a strange problem for a long time with Transcriber 
1.5.1 and wouldn't mind seeing if anyone on the list understands what's 
going on.

When I open a .wav file in Transcriber (recorded at 48/24 on a Zoon 
H4n), the waveform view looks like it's completely blown out, as though 
the gain were set much too high. But the audio sounds fine (note that 
adjusting the gain on the control panel only compresses the view of the 
whole blown-out waveform, it doesn't actually zoom out from it). This is 
a problem for chunking, since a solid line can't be used to anticipate 
breakpoints. Now, when I import the .wav file into Audacity and export 
it as an edited .wav file - even if I haven't actually edited it - the 
waveform suddenly looks fine, and sounds exactly the same as before. 
Does anyone understand what's going on, and maybe know how to fix it?

Thanks very much in advance,

Mark



More information about the Resource-network-linguistic-diversity mailing list