Common Voice Training

Archives DeepSpeech
#1

Hey,

This is my first time working with DeepSpeech and am just going off the documentation and previous posts I have found off this discourse.

I have download the “Common Voice Corpus 6.1” dataset. I have been using the official documentation provided (https://deepspeech.readthedocs.io/en/v0.9.3/) and the DeepSpeech playbook (https://mozilla.github.io/deepspeech-playbook/TRAINING.html). I have followed the steps on a Manjaro (Arch Distro) installation with a 1080 Ti and a AMD Ryzen 7 3700X (8 Core, 16 Thread, 3.600GHz) and have not run into any problems, however, I want to make sure that the command I formulated will get me the fastest possible setup. The command I have created and am currently using is

python3 DeepSpeech.py \
	--train_files deepspeech-data/cv-corpus-6.1-2020-12-11/en/clips/train.csv \
	--dev_files deepspeech-data/cv-corpus-6.1-2020-12-11/en/clips/dev.csv \ 
	--test_files deepspeech-data/cv-corpus-6.1-2020-12-11/en/clips/test.csv \ 
	--train_batch_size 64 \
	--dev_batch_size 64 \
	--test_batch_size 64 \
	--automatic_mixed_precision=True \
	--export_dir deepspeech-data/export \
	--export_tflite \
	--checkpoint_dir deepspeech-data/checkpoint

Thanks in advance

#2

Looks good to me, you can also join us on Mozilla’s Matrix to get realtime help.

#3

Alright, thank you very much. I’ll make sure to join in the situation of which I run into any errors.