Improving accuracy by creating a specific model

[JRMeyer]

>>> jehoshua
[January 21, 2018, 10:21pm]

Have recently installed Deepspeech and run it against a small WAV file -
55 words slash
275 characters, length - 19.968s

The accuracy is not good at all, with an error rate of 43.63%. If I use
a small script to test the same audio with Google Speech Recognition ,
the error rate is only 20%. Still high but better.

Have even cut that WAV down to 9 seconds, yet no change in accuracy.
Initially used the command as per docs

deepspeech models/output_graph.pb my_audio_file.wav models/alphabet.txt

and have added the files lm.binary and trie to the command, and tests
reveal no change.

The audio is English and very clear. I followed the guidlines re
specifications for the WAV and it is the same as the sample WAV's for
Deepspeech.

In regards to these sample model files:

lm.binary slash
trie slash
output_graph.pb

what bearing do they have on accuracy ? Do I need to create a specific
model to improve the accuracy, or 'add to' (train) the sample models ?
There are hundreds of audio files that we have for just one
speaker/person, and we wish to be able to improve the accuracy
substantially.

Just wondering that because this is for a specific purpose, is it better
to create a model for that purpose only, rather than use the existing
(sample) models ?

[This is an archived TTS discussion thread from discourse.mozilla.org/t/improving-accuracy-by-creating-a-specific-model]

[JRMeyer]

>>> jehoshua
[January 21, 2018, 11:00pm]

This tutorial is very good and may be the solution - TUTORIAL : How I
trained a specific french model to control my
robot

[Archived Post]

[JRMeyer]

>>> elpimous_robot
[January 23, 2018, 4:59pm]

Hi. What is your 'specific purpose' ? slash
The answer is difficult :

1/ You have a powerfull PC, and want, in the futur, use your own model
for a complete STT model, why not keep the provided Deepspeech model,
and improve it with your voice and sentences ?! slash
Run a new train :

ex : Feed in your own params !!!

> python -u DeepSpeech.py slash
> --initialize_from_frozen_model
> /home/nvidia/DeepSpeech/data/alfred/results/output_graph0.095.pb slash
> --checkpoint_dir
> /home/nvidia/DeepSpeech/data/alfred/fine_tuning_checkpoints slash
> --train_files /home/nvidia/DeepSpeech/data/alfred/xxx.csv slash
> --dev_files /home/nvidia/DeepSpeech/data/alfred/xxx.csv slash
> --test_files /home/nvidia/DeepSpeech/data/alfred/xxx.csv slash
> --train_batch_size 20 slash
> --dev_batch_size 20 slash
> --test_batch_size 15 slash
> --n_hidden 375 slash
> --epoch 18 slash
> --validation_step 1 slash
> --early_stop True slash
> --earlystop_nsteps 10 slash
> --estop_mean_thresh 0.03 slash
> --estop_std_thresh 0.03 slash
> --dropout_rate 0.12 slash
> --learning_rate 0.001 slash
> --beam_width 500 slash
> --report_count 50 slash
> --lm_weight 5 slash
> --valid_word_count_weight 3 slash
> --decoder_library_path
> /home/nvidia/tensorflow/bazel-bin/native_client/libctc_decoder_with_kenlm.so slash
> --alphabet_config_path /home/nvidia/DeepSpeech/data/alphabet.txt slash
> --lm_binary_path /home/nvidia/DeepSpeech/data/lm.binary slash
> --lm_trie_path /home/nvidia/DeepSpeech/data/trie slash slash

2/Your PC is lower as you expected (they are, LOL): slash
Why not create a homemade model from scratch ?

[Archived Post]

[JRMeyer]

>>> jehoshua
[January 24, 2018, 1:34am]



> 2/Your PC is lower as you expected (they are, LOL): slash
> Why not create a homemade model from scratch ?

Option 2 seems more sensible, as the load of DeepSpeech running on that
computer is unacceptable. I feel adding to the current model will only
put extra load on the computer. These audios are from 44 minutes to 1 hr
18 mins, so the 5 to 10 seconds audio length limitation at present, is
simply not an option.

Has anyone been able to use DeepSpeech on audios of an hour or more ? If
so, how long did it take and what are the specifications of the computer
?

I have even wondered if I should learn to touch type, to create
transcriptions ...seriously.
![:slight_smile:](

[Archived Post]