This is an example project to demonstrate how tooling can be used to generate phonetic descriptions for words using a statistical approach, assuming that a large enough phonetical lexicon exists.
Documentation which was used to generate these examples can be found in the "report" folder.
Inputs:
- manually created example lexicon ("lexicon.lex")
- example list of new words ("hsb.vocab") to generate phonemes for
- see folder "sources/"
Running:
- Build the container using the supplied "Dockerfile"
- see also inline comments
docker build [--build-arg ARCHITECTURE=XXXXX] -t speech_recognition_acoustic_model_training_step1 .-
Build argument "ARCHITECTURE":
- leave out "ARCHITECTURE" to build for x86_64
- specify the following for different architectures:
- Raspberry Pi 0/1: ARCHITECTURE=linux_armv6l
- Raspberry Pi 2/3/4 (32-bit): ARCHITECTURE=linux_armv7l
- Raspberry Pi 3/4 (64-bit): ARCHITECTURE=linux_aarch64
-
Run the example commands to train the lexicon and to predict phonemes for single words or a word list
- see Dockerfile for details
Outputs:
- model ("g2p.fst") trained from the lexicon
- see folder "model/"
- generated lexicon ("hsb_g2p.lex") using the model
- see folder "output/"
-
Dr. Ivan Kraljevski (Fraunhofer Institute for Ceramic Technologies and Systems IKTS, Dresden, Germany)
-
Daniel Sobe (Foundation for the Sorbian people)
See file "LICENSE".