Skip to content
This repository was archived by the owner on Jun 9, 2022. It is now read-only.

Conversation

zamiron
Copy link

@zamiron zamiron commented Feb 16, 2016

Hi Team,

This is new CMN algorithm, that can working "online" and has support cep-normalization.

Best regards,
Zamir

int32 i, j;
mfcc_t cep_cur[cmn->veclen]; // is it correctly?

if (nfr <= 0)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Add curly braces.

@mbait
Copy link
Contributor

mbait commented Feb 16, 2016

Please follow the project's coding style.

@zamiron
Copy link
Author

zamiron commented Feb 16, 2016

Hi Alexander,

These style-errors have been fixed.
Thank you for your comments.

if ( cep_cur[0] > cmn->max[0] )
cmn->max[0] = cep_cur[0] * prob0 + cmn->max[0] * (1.0f-prob0);

mfcc_t e_prob = incep[i][0] / cmn->max[0] * 0.001f;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What does 0.001f mean here. Also please note that mfcc_t can be a fixed point number (int) in fixed point mode, floating point operation is not easy here.

@zamiron
Copy link
Author

zamiron commented Feb 17, 2016

Hi Team!

Test "-cmn adapt" with option "-remove_silence no"
My test result:
With VAD: Percent Word Accuracy = 80.8%
Without VAD: Percent Word Accuracy = 81.1%

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Development

Successfully merging this pull request may close these issues.

3 participants