Forced alignment
·Lets see next process ,download latest Montreal Force Aligner source code from Link :- Montreal Forced Aligner Pre-trained model I used is Spanish you can download it from MFA website. before doing this check whether you have latest python module python 3.x if not installed that first using sudo apt-get install python3 . Also you should have pip package above pip3.x. pip install –upgrade pip 1) Open terminal and change directory to unzipped folder of MFA. 2)Run
bin/mfa_align .
kranti@kranti:~/Montreal-Forced-Aligner1$ bin/mfa_align
usage: mfa_align [-h] [-s SPEAKER_CHARACTERS] [-t TEMP_DIRECTORY]
[-j NUM_JOBS] [-v] [-n] [-c] [-d] [-e] [-i]
corpus_directory dictionary_path acoustic_model_path
output_directory
mfa_align: error: the following arguments are required: corpus_directory, dictionary_path, acoustic_model_path, output_directory
kranti@kranti:~/Montreal-Forced-Aligner1$Such usage information should be printed out. If not change directory and follow third step 3)
cd/path/to/Montreal-Forced-Aligner/thirdpartyThen Run
python3 kaldibinaries.py ~/kaldi/rootkranti@kranti:~/Montreal-Forced-Aligner1/thirdparty$ python3 kaldibinaies.py ~/kaldi/rootNow change directory to Montreal-Forced-Aligner and run following command.
kranti@kranti:~/Montreal-Forced-Aligner-master$ pip install -r requirements.txt
Collecting https://github.com/pyinstaller/pyinstaller/archive/develop.zip (from -r requirements.txt (line 5))
Downloading https://github.com/pyinstaller/pyinstaller/archive/develop.zip
- 10kB 33kB/s^ZIt would download all packages mentioned inside requirements.txt. Now run
freezing/freeze.shthere will be folder created by the name dist change directory to that folder and further change directory to folder named montreal-forced-aligner and change directory to bin this should contain two files mfa_align and mfa_train_and_align . Now run
bin/mfa_align or
bin/mfa_train_and_alignfrom montreal-forced-aligner directory output should be similar to output message mentioned in step 1.With all steps done under Kaldi compile and Montreal Aligner setup now is the time to align any given .wav file and .lab file (which contains transcript of given wav file).
NOTE :> The wav file mentioned should be atleast 16kHz sampling rate.The aligner ignores the files having sampling rate below 16kHz. The features that the aligner extracts are MFCCs dervied from spectra from 0 to 7.8 kHz, so sound files must have Nyquist frequencies higher than 7.8 kHz.To find a sample rate in a Linux system Right click the file, then click “Properties.” The sample rate information may appear in a different tab of the Properties box depending on your Linux distribution and version. If it doesn’t appear in the main tab, look for an “Audio” or “More Info” tab.
Let’s start aligning.
My Home directory has :
| FOLDER | CONTENTS |
|---|---|
| Montreal-Forced-Aligner-master | source code from github |
| MFATestWav | containing .wav and .lab file (corpus path) |
| SpanishDict | contains spanish dictionary(dictionary path) |
| SpanishModel | contains Spanish Model from MFA website in spanish.zip form |
kranti@kranti:~/Montreal-Forced-Aligner-master$ bin/mfa_train_and_align -t ~/output -v ~/MFATestWav ~/SpanishDict/Spanishdict.txt spanish
Setting up corpus information...
Creating dictionary information...
Setting up training data...
Calculating MFCCs...
Calculating CMVN...
Number of speakers in corpus: 1, average number of utterances per speaker: 1.0
Beginning monophone training...
0%| | 0/39 [00:00<?, ?it/s]log-likelihood -83.4514
skipped transitions 140 183
missing data gaussians 6
3%|█▌ | 1/39 [00:00<00:13, 2.91it/s]log-likelihood -81.0823
skipped transitions 140 183
missing data gaussians 6
5%|███▏ | 2/39 [00:00<00:11, 3.21it/s]log-likelihood -81.0169
skipped transitions 140 183
missing data gaussians 6
8%|████▊ | 3/39 [00:00<00:10, 3.46it/s]log-likelihood -80.9504
skipped transitions 140 183
missing data gaussians 6
10%|██████▍ | 4/39 [00:01<00:09, 3.65it/s]log-likelihood -80.3842
skipped transitions 140 183
missing data gaussians 6First Mono-Phone Training.
Second Tri-Phone Training.
Third Speaker-Adapted Tri-Phone Training.
Command is :>
bin/mfa_train_and_align -t ~/output -v ~/path/to/corpus ~/path/to/dict.txt output_directoryYou can also use other command arguments such as -s > use to allot number of characters to use to identify speakers. -t > Temporary directory to use for aligining .(~/Documents/MFA). -j > Number of jobs to use. (default = 3). -v > verbose. -c > Temporary files in ~/Documents/MFA and the output directory will be removed prior to aligning. This is good to use when aligning a new dataset, but it shares a name with a previously aligned dataset. -h > Display help message. -i > ignore. -q > quiet.
USING PRETRAINED MODEL:
bin/mfa_align corpus_directory dictionary_path acoustic_model_path output_directorykranti@kranti:~/Montreal-Forced-Aligner-masters bin/mfa_align -t ~/align -v ~/MFATestHav ~/Spanis
hDict/Spd2.txt ~/SpanishModel/output.zip spanish
Setting up corpus information...
Number of speakers in corpus: 1, average number of utterances per speaker: 1.0
Creating dictionary information...
Setting up training data...
Calculating MFCCs...
Calculating CMVN...
Number of speakers in corpus: 1, average number of utterances per speaker: 1.0
Done with setup.
There were words not found in the dictionary
100%
| 2/2 [oo:oo<oo:oo, 2.43it/s]
Done! Everything took 5.192259311676025 seconds
kranti@kranti:~/Montreal-Forced-Aligner-masters I
Would ou like to abort to fix them? (Y N)noutput.zip contains output acoustic model using mfa_train_and_align .Similarly we can use pre-trained acoustic model.
Now to check output,in Montreal-Forced-Aligner-master under folder named spanish/MFATestWav/1.TextGrid Output of MFA is always in textgrid form ,to check alignment results we use praat. Praat is used for Speech Analysis,Speech Synthesis,Speech Manipulation,Labelling and Segmentation,etc. Our purpose is related to analysis.Lets view output of textgrid. Run :
kranti@kranti:~/Montreal-Forced-Aligner-master$ praat --open 1.TextGrid 1.wavLook you have to include 1.wav file in current directory.