kaldi-developers

Contact owners and managers

1–30 of 396

Both programming and recipes questions and discussion. If you are just starting, perhaps you should consider kaldi-help first.

IMPORTANT! PLEASE GO TO http://kaldi-asr.org/forums.html TO FIND OUT HOW TO JOIN!

0 selected

Travis C, Nickolay Shmyrev3

Jul 11

Kaldi Non-English Features Trouble

Thanks Nickolay, good to know about the pitch feature, it was tough for me to parse out how important

unread,

Kaldi Non-English Features Trouble

Thanks Nickolay, good to know about the pitch feature, it was tough for me to parse out how important

Jul 11

JC, Daniel Povey2

May 3

Missing OBJFILES Entry in Makefile

I suspect what happened is that the Conda release is super out of date. I was not actually aware that

unread,

Missing OBJFILES Entry in Makefile

I suspect what happened is that the Conda release is super out of date. I was not actually aware that

May 3

Jayenthiran Pukuraj

May 2

Need British English Speech Dataset with Mispronunciation-Aware Decoding

HI I'm working on an ASR-based project for my liguistic training targeted toward English language

unread,

Need British English Speech Dataset with Mispronunciation-Aware Decoding

HI I'm working on an ASR-based project for my liguistic training targeted toward English language

May 2

Abraham Nyongesa, … Jayenthiran Pukuraj3

Apr 22

Training on GPU taking very long time

Hi Desh, I've been learning and working with Kaldi for the past two months, and I'm happy to

unread,

Training on GPU taking very long time

Hi Desh, I've been learning and working with Kaldi for the past two months, and I'm happy to

Apr 22

P76111597李熙堃 LEE SHE KUN

Mar 21

Re: [kaldi-developers] TEAMS CALL FOR HELP - Kaldi Mini-Librespeech development

What do you want to know? On Fri, Mar 21, 2025 at 2:19 PM Jayenthiran Pukuraj <jaya...@gmail.com

unread,

Re: [kaldi-developers] TEAMS CALL FOR HELP - Kaldi Mini-Librespeech development

What do you want to know? On Fri, Mar 21, 2025 at 2:19 PM Jayenthiran Pukuraj <jaya...@gmail.com

Mar 21

Abraham Nyongesa2

12/2/24

Re: [kaldi-developers] Digest for kaldi-developers@googlegroups.com - 1 update in 1 topic

Create text. That's the transcription On Mon, 2 Dec 2024, 4:46 pm Abraham Nyongesa, <abraham.

unread,

Re: [kaldi-developers] Digest for kaldi-developers@googlegroups.com - 1 update in 1 topic

Create text. That's the transcription On Mon, 2 Dec 2024, 4:46 pm Abraham Nyongesa, <abraham.

12/2/24

Dimitri TCHAHEU TCHAHEU

10/9/24

How to get phone statistics

Hello every body. How can we obtain the number of occurence of each phone with kaldi. I have trained

unread,

How to get phone statistics

Hello every body. How can we obtain the number of occurence of each phone with kaldi. I have trained

10/9/24

Nikhil, Dimitri TCHAHEU TCHAHEU4

10/9/24

MFCC extraction jobs failed

Check the number of speaker if it is lower than the number of job, this can be error source Le jeudi

unread,

MFCC extraction jobs failed

Check the number of speaker if it is lower than the number of job, this can be error source Le jeudi

10/9/24

Ravindra Jayasundara, Dimitri TCHAHEU TCHAHEU2

10/9/24

Librispeech Example Running

It seem that you have not provide the segment file, check the kaldi data preparation here Le vendredi

unread,

Librispeech Example Running

It seem that you have not provide the segment file, check the kaldi data preparation here Le vendredi

10/9/24

Gisel Caicedo, … Desh Raj4

10/4/24

Kaldi CPU Only

You can check out the distil-whisper model on faster-whisper ( https://github.com/SYSTRAN/faster-

unread,

Kaldi CPU Only

You can check out the distil-whisper model on faster-whisper ( https://github.com/SYSTRAN/faster-

10/4/24

Dimitri TCHAHEU TCHAHEU

7/23/24

Extra question

Hello, my name is Tchaheu Tchaheu Dimitri. I am training an HMM acoustic model on a local Cameroonian

unread,

Extra question

Hello, my name is Tchaheu Tchaheu Dimitri. I am training an HMM acoustic model on a local Cameroonian

7/23/24

Max Lvov, … Aditya Parikh5

4/1/24

SSL features instead of MFCC in Kaldi?

Hi Ondrej, Thanks for mentioning the paper. Is there any codebase available to replicate the

unread,

SSL features instead of MFCC in Kaldi?

Hi Ondrej, Thanks for mentioning the paper. Is there any codebase available to replicate the

4/1/24

박상훈

9/7/23

Do you have Tips when you have to recognize one word

Hi members I use kaldi very hard and when I make LM model, I use N-Gram so, to recognize one word (

unread,

Do you have Tips when you have to recognize one word

Hi members I use kaldi very hard and when I make LM model, I use N-Gram so, to recognize one word (

9/7/23

wenqian bao, Daniel Povey2

8/15/23

Some question about kaldi_io and .ark

Maybe they are different in very insignificant digits that are not printed in text mode of copy-

unread,

Some question about kaldi_io and .ark

Maybe they are different in very insignificant digits that are not printed in text mode of copy-

8/15/23

natan...@gmail.com, Jan Yenda Trmal2

5/7/23

FLAC vs PCM

PCM.and flac ať the same sampling frequency and sample bit width should give comparable results. That

unread,

FLAC vs PCM

PCM.and flac ať the same sampling frequency and sample bit width should give comparable results. That

5/7/23

wenqian bao

4/22/23

If I use the pretrained model but the input-dim isn't match ,how can i adjust input-dim of my wavs?

If I use the pretrained model but the input-dim isn't match ,how can i adjust input-dim of my

unread,

If I use the pretrained model but the input-dim isn't match ,how can i adjust input-dim of my wavs?

If I use the pretrained model but the input-dim isn't match ,how can i adjust input-dim of my

4/22/23

wenqian bao

4/21/23

About gop recipe

I meet this error when I use chinese dataset to run gop recipe，how can I solve it? ERROR (nnet3-

unread,

About gop recipe

I meet this error when I use chinese dataset to run gop recipe，how can I solve it? ERROR (nnet3-

4/21/23

Reethu Chowdary

3/23/23

Running KWS librispeech locally for few words

Hi, I do not have any OOV words. I just want to use librispeech data and do KWS. The model generated

unread,

Running KWS librispeech locally for few words

Hi, I do not have any OOV words. I just want to use librispeech data and do KWS. The model generated

3/23/23

wenqian bao, Yenda2

3/6/23

Some question about recipe gop

I think the discriminatively trained models are generally not giving the right probabilities/

unread,

Some question about recipe gop

I think the discriminatively trained models are generally not giving the right probabilities/

3/6/23

Abraham Nyongesa, Desh Raj2

3/6/23

SPEECH SPERATION MODULE

I don't think there are any speech separation recipes in Kaldi. You may want to check out

unread,

SPEECH SPERATION MODULE

I don't think there are any speech separation recipes in Kaldi. You may want to check out

3/6/23

wenqian bao

3/5/23

I want to know how to learn gop with kaldi

Hi, I need to make a research with kaldi-gop,but I'm a beginner，can you give me a introduction on

unread,

I want to know how to learn gop with kaldi

Hi, I need to make a research with kaldi-gop,but I'm a beginner，can you give me a introduction on

3/5/23

Abraham Nyongesa

3/1/23

Re: [kaldi-developers] Digest for kaldi-developers@googlegroups.com - 2 updates in 1 topic

Thank you On Wed, Mar 1, 2023, 8:18 PM <kaldi-de...@googlegroups.com> wrote: kaldi-

unread,

Re: [kaldi-developers] Digest for kaldi-developers@googlegroups.com - 2 updates in 1 topic

Thank you On Wed, Mar 1, 2023, 8:18 PM <kaldi-de...@googlegroups.com> wrote: kaldi-

3/1/23

ishtiaqu...@gmail.com

11/25/22

Chain2 TDNN Training error

Dear Everyone, I am facing error while doing the TDNN whenever i am trying to run a script i am

unread,

Chain2 TDNN Training error

Dear Everyone, I am facing error while doing the TDNN whenever i am trying to run a script i am

11/25/22

marc....@protonmail.com, … Daniel Povey5

11/21/22

Union of n-gram FSTs failing at decode time...

Thanks Dan! It seems to be an easy and much safer way to do it! That'll work I think! Dan Povey

unread,

Union of n-gram FSTs failing at decode time...

Thanks Dan! It seems to be an easy and much safer way to do it! That'll work I think! Dan Povey

11/21/22

Abraham Nyongesa

11/17/22

Re: [kaldi-developers] Digest for kaldi-developers@googlegroups.com - 2 updates in 1 topic

Thank you On Thu, Nov 17, 2022, 8:17 PM <kaldi-de...@googlegroups.com> wrote: kaldi-

unread,

Re: [kaldi-developers] Digest for kaldi-developers@googlegroups.com - 2 updates in 1 topic

Thank you On Thu, Nov 17, 2022, 8:17 PM <kaldi-de...@googlegroups.com> wrote: kaldi-

11/17/22

Abraham Nyongesa, Jan Yenda Trmal2

11/17/22

Grid engine

there is basic walkthrough here http://kaldi-asr.org/doc/queue.html On Thu, Nov 17, 2022 at 10:03 AM

unread,

Grid engine

there is basic walkthrough here http://kaldi-asr.org/doc/queue.html On Thu, Nov 17, 2022 at 10:03 AM

11/17/22

marc....@protonmail.com3

10/7/22

Nnet3 raw DNN online forward pass...

I got it working. The issue was that a chunk of feature vectors does not have enough context to

unread,

Nnet3 raw DNN online forward pass...

I got it working. The issue was that a chunk of feature vectors does not have enough context to

10/7/22

Abraham Nyongesa

10/7/22

VOXCELEB DATASET PREPARATION

I am downloading audios from videos url and timestamps given in voxceleb dataset. In every url folder

unread,

VOXCELEB DATASET PREPARATION

I am downloading audios from videos url and timestamps given in voxceleb dataset. In every url folder

10/7/22

Abraham Nyongesa, … Desh Raj7

10/4/22

voxceleb recipe

If you are facing issues in Lhotse installation, you can create a new issue on the repo with details

unread,

voxceleb recipe

If you are facing issues in Lhotse installation, you can create a new issue on the repo with details

10/4/22

Jan Yenda Trmal

9/1/22

Re: [kaldi-developers] Convert phone level .ctms to word level ctms?

there is not. ctm is usually a final output (or in some cases the input). There are no tools in kaldi

unread,

Re: [kaldi-developers] Convert phone level .ctms to word level ctms?

there is not. ctm is usually a final output (or in some cases the input). There are no tools in kaldi

9/1/22