| Both sides previous revision
Previous revision
Next revision
|
Previous revision
|
forced-aligner [2018/03/22 14:56]
mganzeboom |
forced-aligner [2019/06/20 09:10] (current)
mhulsbosch |
| |
| **Authors**: Linde Kuijpers (student assistent), Mario Ganzeboom (PhD | **Authors**: Linde Kuijpers (student assistent), Mario Ganzeboom (PhD |
| student, m.ganzeboom@let.ru.nl) \\ | student, m.ganzeboom@let.ru.nl), Xing Wei (PHD, X.Wei@let.ru.nl) \\ |
| **Last changes in code**: 16-03-2018 \\ | **Last changes in code**: 14-04-2019 \\ |
| **Last changes in readme**: 22-03-2018 \\ | **Last changes in readme**: 19-06-2019 \\ |
| **Current location**: /vol/tensusers/mganzeboom/clst-asr_forced-alignment | **Current location**: /vol/tensusers/xwei/clst-asr_forced-aligner/ |
| |
| **Function**: forced alignment of speech recordings using NNet2 online acoustic models trained with the Kaldi ASR toolkit (http://kaldi-asr.org). Currently, this tool includes acoustic models for Dutch trained on the Spoken Dutch Corpus (SDC, or 'CGN' in Dutch) by the Open-source Nederlandse spraakherkenning project ([[http://www.opensource-spraakherkenning.nl]]). | **Function**: forced alignment of speech recordings using NNet2 online acoustic models trained with the Kaldi ASR toolkit (http://kaldi-asr.org). Currently, this tool includes acoustic models for Dutch trained on the Spoken Dutch Corpus (SDC, or 'CGN' in Dutch) by the Open-source Nederlandse spraakherkenning project ([[http://www.opensource-spraakherkenning.nl]]). |
| **How to run the script**: | **How to run the script**: |
| - Login to one of the ponies (do not use (the old) applejack because of an older CUDA version!). | - Login to one of the ponies (do not use (the old) applejack because of an older CUDA version!). |
| - Run from your Home directory: /vol/tensusers/mganzeboom/clst-asr-forced-aligner/run.sh <absolute-path-to-directory-with-recordings>. The script will create a directory in your Home dir and copy the default config. | - Run from your Home directory: /vol/tensusers/xwei/clst-asr_forced-aligner/run.sh <absolute-path-to-directory-with-recordings>. The script will create a directory in your Home dir and copy the default config and lexicon file. |
| - Open ~/clst-asr-fa/align_config.rc with your favourite editor and change the configuration settings to your liking (the defaults are fine on average). | - Open ~/clst-asr-fa/align_config.rc with your favourite editor and change the configuration settings to your liking (the defaults are fine on average). Please make sure the tier name in this config file is consistent with with your *.tg files!!! |
| - Run step 2 once again and a job will be added to the Slurm queue manager starting the forced alignment of the provided directory. The logs of this job can be found in ~/clst-asr-fa/slurm-logs/slurm-<job-id>.log. Provide multiple input directories at the command line to queue multiple jobs at once. | - Run step 2 once again and a job will be added to the Slurm queue manager starting the forced alignment of the provided directory. The logs of this job can be found in ~/clst-asr-fa/slurm-logs/slurm-<job-id>.out. Provide multiple input directories at the command line to queue multiple jobs at once. |
| - The force alignment logs can be found at <absolute-path-to-directory-with-recordings>/logs when all Slurm jobs have completed. | - The force alignment logs can be found at <absolute-path-to-directory-with-recordings>/logs when all Slurm jobs have completed. |
| - It may occur that certain words from the transcriptions are not in the lexicon provided with the acoustic models. Run the script /vol/tensusers/mganzeboom/clst-asr-forced-aligner/list-missing-words.sh <absolute-path-to-directory-with-recordings> to print a list of these words and their corresponding transcription files to the command line. It is unfortunately not yet possible for you as a user to add new words to the lexicon. In that case you first need to copy the forced aligner (i.e. /vol/tensusers/mganzeboom/clst-asr-forced-aligner) to your directory on ponyland. | - It may occur that certain words from the transcriptions are not in the lexicon provided with the acoustic models. Run the script /vol/tensusers/xwei/clst-asr_forced-aligner/list-missing-words.sh <absolute-path-to-directory-with-recordings> to print a list of these words and their corresponding transcription files to the command line. You could then add a phonemic transcription of these words to your custom lexicon file in ~/clst-asr-fa/lexicon.txt. It is recommended to base these new transcriptions on parts of already existing ones. Afterwards, rerun the script from step 4. |
| | - All the audio files and *.tg files will be moved to a folder named source_files (path is <absolute-path-to-directory-with-recordings>/source_files), alongside with a folder named results, which was created to save all the generated *_aligned.TextGrid with alignment results. The log, source_files and results folder are created automatically. |
| | - After process finished, please check the log file (path:~/clst-asr-fa/slurm-logs/) if there are some errors. |
| | |
| | For the forced aligner for English: |
| | - use /vol/tensusers/xwei/clst-eng_forced-aligner/run.sh in step 2 |
| | - the directory that is created in you home dir is ~/clst-eng-fa |
| | |
| | Login to see discussion on improvements below. |
| |
| ~~DISCUSSION~~ | ~~DISCUSSION~~ |
| |
