Awesome

forced-alignment

Tools for forced alignment. Fork of p2fa.

<br> What's here ----------- 1. [`create_transcripts.py`](https://github.com/myedibleenso/forced-alignment/blob/master/transcripts/create_transcripts.py)

An original script for automatic generation of **htk**-compatible transcripts from some set of source sentences

2. pronunciator.py

Pronunciation estimator for unknown utterances (requires a pronunciation dictionary). &nbsp;A copy of the [CMU pronouncing dictionary](http://www.speech.cs.cmu.edu/cgi-bin/cmudict) is provided [here](https://github.com/myedibleenso/forced-alignment/blob/master/pronunciation/cmudict).  This script is able to generate .csv and LaTeX-formatted table of compositional statistics for a specified transcript or set of sentences (i.e. average length, \# syllables, syllables per word, unique phones, etc).<p>

4. align.py

Minimally modified **p2fa** that resolves deprecated SoX call<p>

5. align_batch.py

Script to batch process multiple wav files with minimally modified **p2fa**.

6. align_new.py

Revised **p2fa** with several new features

7. splice.praat

Praat script for TextGrid-based slicing of wav files.  &nbsp;Credit and thanks given to [Jessamyn Schertz](http://www.u.arizona.edu/~jschertz/index.shtml)<p>

<br>

Setup

Install htk 3.4. Careful about this point as v3.4.1 will not work.

Mac users experiencing problems should look here.

<br> Other stuff -----------

File naming convention:

<GENDER><GENDER-ID>_HS_<NATIVE or NON-NATIVE (N or NN)>_<SPEED (NORM or FAST)>_<LIST>-<SENTENCE NUM (1-10)>

MORE RECORDINGS: http://www.voiptroubleshooter.com/open_speech/american.html