.. _adapt_acoustic_model:

Adapt acoustic model to new data ``(mfa adapt)``
================================================

.. deprecated:: 3.4

   Beginning with MFA 4.0 (currently scheduled for end of 2026), the default signature of ``mfa adapt`` will be changed to the signature of :ref:`adapt_acoustic_model_hf` and the signature for use with legacy models will be deprecated (this version of the command will become ``mfa adapt_legacy``. The ``mfa adapt_legacy`` command will be removed in MFA 5.0 (currently scheduled for some time in 2027). See :ref:`mfa_model_versions` for more details on the difference.

MFA can adapt pretrained :term:`acoustic models` to a new dataset.  MFA will first align the dataset using the pretrained model, and then update the acoustic model's GMM means with those generated by the data.  See :kaldi_steps:`train_map` for the Kaldi script this functionality corresponds to.  As part of the adaptation process, MFA can generate final alignments and export these files if an output directory is specified in the command.


.. note::

   You can use manual or verified reference alignments in adaptation to bypass the initial round of alignment for some or
   all files and these alignments will be used for adjusting the acoustic model parameters.  See :ref:`reference_alignment_format`
   for more information on how to include these alignments.


.. seealso::

   See :xref:`mfa_adaptation_scripts` for reference files and CLI commands that are have been used for evaluating performance of adaptation.

Command reference
-----------------

.. click:: montreal_forced_aligner.command_line.adapt:adapt_model_cli
   :prog: mfa adapt
   :nested: full

Configuration reference
-----------------------

- :ref:`configuration_global`
- :ref:`configuration_adapting`

API reference
-------------

- :class:`~montreal_forced_aligner.alignment.AdaptingAligner`

.. _adapt_acoustic_model_hf:

Adapt MFA model on Hugging Face to new data ``(mfa adapt_hf)``
==============================================================

MFA can adapt pretrained :term:`acoustic models` to a new dataset.  MFA will first align the dataset using the pretrained model, and then update the acoustic model's GMM means with those generated by the data.  See :kaldi_steps:`train_map` for the Kaldi script this functionality corresponds to.  As part of the adaptation process, MFA can generate final alignments and export these files if an output directory is specified in the command.


.. deprecated:: 3.4

   Beginning with MFA 4.0, the ``mfa adapt_hf`` command will become the default ``mfa adapt`` command and the legacy version of the command changed to ``mfa adapt_legacy``. See :ref:`mfa_model_versions` for more details on the difference.

.. note::

   You can use manual or verified reference alignments in adaptation to bypass the initial round of alignment for some or
   all files and these alignments will be used for adjusting the acoustic model parameters.  See :ref:`reference_alignment_format`
   for more information on how to include these alignments.


.. seealso::

   See :xref:`mfa_adaptation_scripts` for reference files and CLI commands that are have been used for evaluating performance of adaptation.

Command reference
-----------------

.. click:: montreal_forced_aligner.command_line.adapt:adapt_model_cli
   :prog: mfa adapt
   :nested: full

Configuration reference
-----------------------

- :ref:`configuration_global`
- :ref:`configuration_adapting`

API reference
-------------

- :class:`~montreal_forced_aligner.alignment.AdaptingAligner`