The mfa command line utility has several subcommands, which are listed below grouped by general domain.

Forced Alignment

Command Description Link
align Perform forced alignment with a pretrained model Align using pretrained models
train Train an acoustic model and export resulting alignment Align using only the data set
adapt Adapt a pretrained acoustic model on a new dataset Adapt pretrained models to new dataset
validate Validate a corpus to ensure there are no issues with the data format Validating data
train_dictionary Estimate pronunciation probabilities from aligning a corpus Modeling pronunciation probabilities


Command Description Link
transcribe Generate transcriptions using an acoustic model, dictionary, and language model Running the transcriber
train_lm Train a language model from a text corpus or from an existing language model Training language models

Corpus creation

Command Description Link
create_segments Use voice activity detection to create segments Create segments
train_ivector Train an ivector extractor for speaker classification Training an ivector extractor
classify_speakers Use ivector extractor to classify files or cluster them Speaker classification
annotator Run a GUI annotator program for editing and managing corpora Annotator

Other utilities

Command Description Link
download Download a model trained by MFA developers Pretrained models
thirdparty Download and validate new third party binaries Installation


Command Description Link
g2p Use a G2P model to generate a pronunciation dictionary Generating a dictionary
train_g2p Train a G2P model from a pronunciation dictionary Training a new G2P model