[LATEST] ValueError: Scorer initialization failed with error code 0x2005

DeepSpeech
,
#1

Hello,

So total noob here, I made a Scorer as per https://mozilla.github.io/deepspeech-playbook/SCORER.html. However I am getting the error ValueError: Scorer initialization failed with error code 0x2005. Am I kinda on the right path :stuck_out_tongue_winking_eye:

ENV: Docker (latest)
GPU: 1070 working as expected
OS: ubuntu 20.04
CONTEXT: Tring to Transcribe Public Safety Radio Audio

COMMAND:

python -u DeepSpeech.py \
  --alphabet_config_path deepspeech-data/alphabet.txt \
  --train_files deepspeech-data/clips/train2.csv \
  --dev_files deepspeech-data/clips/dev2.csv \
  --test_files deepspeech-data/clips/test2.csv \
  --train_batch_size 25 \
  --test_batch_size 11 \
  --dev_batch_size 25 \
  --checkpoint_dir /checkpoint \
  --export_dir=/model \
  --use_allow_growth true \
  --n_hidden 375 \
  --epochs 50 \
  --early_stop true \
  --es_epochs 10 \
  --dropout_rate 0.22 \
  --learning_rate 0.00095 \
  --report_count 100 \
  --audio_sample_rate 48000
  --scorer_path deepseech-data/radio.scorer

SCORER: https://t.rt6.us/lTJ/radio.scorer
https://t.rt6.us/zsQ/vocab-500000.txt
https://t.rt6.us/1lR/vocab.txt
LM-BINARY: https://t.rt6.us/vcN/lm.binary
ACTION: TRAINING (STARTUP)

  raise ValueError('Scorer initialization failed with error code 0x{:X}'.format(err))
ValueError: Scorer initialization failed with error code 0x2005
root@cef8999ee541:/DeepSpeech# deepspeech-data/run.sh
Traceback (most recent call last):
  File "DeepSpeech.py", line 12, in <module>
    ds_train.run_script()
  File "/DeepSpeech/training/deepspeech_training/train.py", line 1033, in run_script
    absl.app.run(main)
  File "/usr/local/lib/python3.6/dist-packages/absl/app.py", line 300, in run
    _run_main(main, args)
  File "/usr/local/lib/python3.6/dist-packages/absl/app.py", line 251, in _run_main
    sys.exit(main(argv))
  File "/DeepSpeech/training/deepspeech_training/train.py", line 998, in main
    early_training_checks()
  File "/DeepSpeech/training/deepspeech_training/train.py", line 983, in early_training_checks
    FLAGS.scorer_path, Config.alphabet)
  File "/usr/local/lib/python3.6/dist-packages/ds_ctcdecoder/__init__.py", line 36, in __init__
    raise ValueError('Scorer initialization failed with error code 0x{:X}'.format(err))
ValueError: Scorer initialization failed with error code 0x2005

head -n 10 radio.scorer

max@josh-Alienware-15-R4:~$ head -n 10 /opt/data/radio.scorer
mmap lm http://kheafield.com/code format version 5
�?������?eg>�L�d��i�ѵs$9��*Thc�JItT�m�V&Y'p��0�Isp~!�9�u��_y���a����K� 2����TϿ�&���C�
                                                                                                      �D�I     :����S :�XrP
�+!1C$R�"5�`B����pr�[�/����٘?�1�ܙw����������e��F��q���h�h���#�f���m�����l{��w�4���1�
��'4* �2Z�:�PI6}Y���v�:��E��0����/ՙ����ܸ�M�u�β��[`�h�|��
                                                                          t���w�=
#2

You incorrectly specified scorer path. Your file is in /opt/data/radio.scorer while in the command line you wrote deepseech-data/radio.scorer. Try to use /opt/data/radio.scorer in command line too.

#3

Hey @nshmyrev,

Thanks for the quick reply!
Im afraid its mapped in docker ‘-v /opt/data:/DeepSpeech/deepspeech-data’

so the path is right in the container

#4

The error means path is wrong, you still need to verify. At least specify an absolute path, the relative path might be wrong.

#5

I promise its there :laughing:

image

#6

My dear, please specify an absolute path /DeepSpeech/deepspeech-data/radio.scorer in the command line.

#7

Just did, I promised

image
image3414×389 34.6 KB

#8

You have deepseech-data instead of deepspeech-data on the screen.

#9

You are 100% right. That did it. Odd relative doesn’t work. Cheers!