Speech:Exps 0063

Description
Author: Tyler Martin Date: (Started) 3/31/2013

Purpose: To test running a mini train for the first time

Details: Followed the guide [Here]. Was able to get the train to run. Continued with language model and decode

Results After class on 3/27/13 where we discussed that there was errors associated with tiny train, I tried the model and decode on my old experiment 0031 but it had errors in scoring. This experiment was created as a result of that.

Following the updated guide, I created this experiment following it once again creating it with no problems. Instead of updating the dictionary with missing words, I copied the dictionary from Eric's working experiment 0058. This was suggested by members of my group. The train ran and completed without any critical errors. Moving onto following the language model guide was very straightforward.

The decode process had one error as there was missing reference files. Using the uniq tool suggested in the guide fixed this error but came back with a second error. There was a duplication in one of the transcripts. So following the guide for removing the duplication, I opened VI and searched for the line above. After deleting one of the lines I tried to run the score again and it completed producing the table below: