MFA Users

Contact owners and managers

1–30 of 117

Thanks for using the Montreal Forced Aligner!

In order for us to help answer questions and address issues as fast as possible, it's helpful to include the following information when you post about an issue encountered:

Full output from when you executed one of our binaries to when it went wrong (this includes status messages as it completes various aspects of set up and alignment, as well as the stack trace that Python outputs)
Any steps that you might have done to try to fix the problem
If possible, find a small subset of the data that replicates the issue and either attach here or send to Michael McAuliffe (michael.e.mcauliffe@gmail.com) to help speed up debugging

In the current version of the aligner (1.0.0), the aligner automatically detects when something has gone wrong and know to force using the clean flag, so previous workarounds using that flag should be unnecessary.

0 selected

caitlin....@gmail.com, michael.e...@gmail.com3

Apr 11

module not found - can't download any acoustic models or dictionaries

(base) C:\Users\55151349>conda activate aligner 1st option (aligner) C:\Users\55151349>conda

unread,

module not found - can't download any acoustic models or dictionaries

(base) C:\Users\55151349>conda activate aligner 1st option (aligner) C:\Users\55151349>conda

Apr 11

11/24/23

Option to turn off allophonic detail in MFA?

Hi all, I've previously used an earlier version of MFA with the ARPABET US English dictionary,

unread,

Option to turn off allophonic detail in MFA?

Hi all, I've previously used an earlier version of MFA with the ARPABET US English dictionary,

11/24/23

Leandro Graciá Gil, Eleanor Chodroff3

8/15/23

MFA vs Epitran phoneme sets

Hi Eleanor, Thank you very much for your response about the Common Voice models. I've been taking

unread,

MFA vs Epitran phoneme sets

Hi Eleanor, Thank you very much for your response about the Common Voice models. I've been taking

8/15/23

Marie Hansen, … Thea Knowles3

8/8/23

Use other tier than tier 1 in a textgrid for alignment

I used a little praat script to get my textgrids into the proper format, then fed those revised

unread,

Use other tier than tier 1 in a textgrid for alignment

I used a little praat script to get my textgrids into the proper format, then fed those revised

8/8/23

Hovsep Deovletian, Thea Knowles2

8/8/23

Is MFA confused by corpus of multiple speakers but with one-to-many ratio?

I was running into speaker adaptation issues as well, and realized that the standard MFA align

unread,

Is MFA confused by corpus of multiple speakers but with one-to-many ratio?

I was running into speaker adaptation issues as well, and realized that the standard MFA align

8/8/23

8/8/23

mfa align --speaker_characters results in Key Error

Update: while -s still doesn't work as expected, I did figure out a workaround for correctly

unread,

mfa align --speaker_characters results in Key Error

Update: while -s still doesn't work as expected, I did figure out a workaround for correctly

8/8/23

Santanu Dasgupta

9/5/22

Phoneme Boundaries

Hi: How do I get the phoneme boundaries when the run the mfa align command. In the output now I only

unread,

Phoneme Boundaries

Hi: How do I get the phoneme boundaries when the run the mfa align command. In the output now I only

9/5/22

8/22/22

PronunciationAcousticMismatchError

Hi all, I have trained my own acoustic model on Librispeech. However when I am trying to align some

unread,

PronunciationAcousticMismatchError

Hi all, I have trained my own acoustic model on Librispeech. However when I am trying to align some

8/22/22

Brian Macwhinney, michael.e...@gmail.com2

6/17/22

Adding out of vocabulary words

Hi Brian, If you upgrade to the latest version, those github_token errors will be resolved.

unread,

Adding out of vocabulary words

Hi Brian, If you upgrade to the latest version, those github_token errors will be resolved.

6/17/22

li...@metaserve.org

6/13/22

Dictionary format problem when using

I'm using MFA 2.0.0rc7, and I'm training a g2p model for Shoshoni then creating a dictionary

unread,

Dictionary format problem when using

I'm using MFA 2.0.0rc7, and I'm training a g2p model for Shoshoni then creating a dictionary

6/13/22

6/3/22

mfa model save dictionary error in 2.0.0rc7

I can confirm that this is resolved in 2.0.0rc8! Jeff Mielke On Tuesday, May 31, 2022 at 12:32:58 PM

unread,

mfa model save dictionary error in 2.0.0rc7

I can confirm that this is resolved in 2.0.0rc8! Jeff Mielke On Tuesday, May 31, 2022 at 12:32:58 PM

6/3/22

5/7/22

set --num_jobs not working during MFA training

I trained acustic model using command like below mfa train --num_jobs 200 -v -t $exp_dir/tmp $

unread,

set --num_jobs not working during MFA training

I trained acustic model using command like below mfa train --num_jobs 200 -v -t $exp_dir/tmp $

5/7/22

4/21/22

Expected language model format

I'm trying to use mfa transcribe. The documentation for this command states that

unread,

Expected language model format

I'm trying to use mfa transcribe. The documentation for this command states that

4/21/22

caitlin....@gmail.com

4/14/22

v2 dictionary edits from oovs_found

I used to do straightforward dictionary edits with v1, just by adding the transcriptions of the words

unread,

v2 dictionary edits from oovs_found

I used to do straightforward dictionary edits with v1, just by adding the transcriptions of the words

4/14/22

caitlin....@gmail.com, R Burn3

4/14/22

mfa align ZeroDivisionError: division by zero

I don't know if you fixed this by now but I've just installed on a different user profile on

unread,

mfa align ZeroDivisionError: division by zero

I don't know if you fixed this by now but I've just installed on a different user profile on

4/14/22

3/24/22

`mfa adapt` does not do GMM or triphone training, same output as `mfa align`

Hi everyone, I'm having an issue where `mfa align` and `mfa adapt` produce the same output. I

unread,

`mfa adapt` does not do GMM or triphone training, same output as `mfa align`

Hi everyone, I'm having an issue where `mfa align` and `mfa adapt` produce the same output. I

3/24/22

3/24/22

Running MFA with speaker adaptation off?

Issue is resolved, I run 'mfa align' to align without SAT, and 'mfa train' to align

unread,

Running MFA with speaker adaptation off?

Issue is resolved, I run 'mfa align' to align without SAT, and 'mfa train' to align

3/24/22

3/2/22

Error when training new model

Hello, We are currently trying to create an acoustic model using our text grids. We have been

unread,

Error when training new model

Hello, We are currently trying to create an acoustic model using our text grids. We have been

3/2/22

3/2/22

non-speech annotations in version 2.0.0rc3

Hi. I'm using mfa version 2.0.0rc3 to align with a transcript that includes {LG} and {SL}.

unread,

non-speech annotations in version 2.0.0rc3

Hi. I'm using mfa version 2.0.0rc3 to align with a transcript that includes {LG} and {SL}.

3/2/22

kpogof...@gmail.com

2/23/22

MFA Newbie!! Help Needed!!

Hello all, I am new to MFA and I need your help figuring out some issues I am having. I am trying to

unread,

MFA Newbie!! Help Needed!!

Hello all, I am new to MFA and I need your help figuring out some issues I am having. I am trying to

2/23/22

2/7/22

v2.0 vs v1.0.1 LJspeech alignment

Hi, I am observing that the phoneme alignments from mfa v1.0.1 are more accurate than v2.0 for

unread,

v2.0 vs v1.0.1 LJspeech alignment

Hi, I am observing that the phoneme alignments from mfa v1.0.1 are more accurate than v2.0 for

2/7/22

1/13/22

Error running Kaldi binaries

Hello, I'm running into this issue when trying to run the mfa with a small folder of recordings

unread,

Error running Kaldi binaries

Hello, I'm running into this issue when trying to run the mfa with a small folder of recordings

1/13/22

Jason Bohland, … michael.e...@gmail.com5

1/7/22

Any record of alignment quality

Gotcha, vowel acoustic quality will likely not get detected by what Kaldi outputs with the default

unread,

Any record of alignment quality

Gotcha, vowel acoustic quality will likely not get detected by what Kaldi outputs with the default

1/7/22

1/6/22

File structure for train_lm and transcribe?

Hi! The documentation does a great job of describing the expected file structure for training an

unread,

File structure for train_lm and transcribe?

Hi! The documentation does a great job of describing the expected file structure for training an

1/6/22

Daniel Wolf, michael.e...@gmail.com3

1/6/22

MFA 2.0.0b11 on Conda?

I'm in a Docker image, where Conda environments are a bit tricky. I ended up specifying the

unread,

MFA 2.0.0b11 on Conda?

I'm in a Docker image, where Conda environments are a bit tricky. I ended up specifying the

1/6/22

Daniel Wolf, michael.e...@gmail.com3

1/3/22

How to enable CUDA for conda-forge MFA?

Thank you for your answer! If none of the Kaldi commands used by MFA use CUDA-optimized matrix

unread,

How to enable CUDA for conda-forge MFA?

Thank you for your answer! If none of the Kaldi commands used by MFA use CUDA-optimized matrix

1/3/22

Dan Wells, michael.e...@gmail.com3

12/15/21

Implementation of per-speaker dictionaries

Thanks very much for the detailed description! I think I have a handle on how to implement this now.

unread,

Implementation of per-speaker dictionaries

Thanks very much for the detailed description! I think I have a handle on how to implement this now.

12/15/21

Marie Hansen, michael.e...@gmail.com3

12/11/21

English acoustic model for previous MFA version (before recent update)

Here's the link for the english acoustic model: https://github.com/MontrealCorpusTools/mfa-models

unread,

English acoustic model for previous MFA version (before recent update)

Here's the link for the english acoustic model: https://github.com/MontrealCorpusTools/mfa-models

12/11/21

12/2/21

Running mfa validate -j 36 only uses one CPU

A few more details: the behavior described pertains to the phase after logging “Setting up training

unread,

Running mfa validate -j 36 only uses one CPU

A few more details: the behavior described pertains to the phase after logging “Setting up training

12/2/21

Daniel Batten, michael.e...@gmail.com3

11/20/21

"attempted relative import above top-level package"

That worked! Thank you. The way you phrased it makes perfect sense and got me to persist and correct

unread,

"attempted relative import above top-level package"

That worked! Thank you. The way you phrased it makes perfect sense and got me to persist and correct

11/20/21

Search

Clear search

Close search

Google apps

Main menu