Welcome, Guest: Register On Nairaland / LOGIN! / Trending / Recent / New
Stats: 3,143,405 members, 7,781,172 topics. Date: Friday, 29 March 2024 at 10:12 AM

Opinions On Speech Recognition Applying Machine Learning Algorithms - Programming - Nairaland

Nairaland Forum / Science/Technology / Programming / Opinions On Speech Recognition Applying Machine Learning Algorithms (999 Views)

What Are Some Machine Learning Algorithms That You Should Always Have A Strong U / Machine Learning Algorithms / Artificial Intelligence And Machine Learning Group (2) (3) (4)

(1) (Reply) (Go Down)

Opinions On Speech Recognition Applying Machine Learning Algorithms by Gideonbabs: 8:44pm On May 11, 2018
Hi Guys, glad i came across this while i was perusing the internet, i read through the threads and i feel in good hands/company. I'm undergoing a research on Speech Recognition. Specifically i'm trying to apply Machine learning Algorithms to speech recognition of Native Nigerian Languages. I am new to the concept of Machine learning, and i have undergone preliminary research on this area. Having read good number of literature i have an idea of what people are doing relating to my work. i know some gurus are here, and i will appreciate it if we can throw light on this. I just stated the research, so I AM CURIOUS TO KNOW IF MY IDEA IS POSSIBLE,THE CHALLENGES I HAVE TO TACKLE TO MAKE IT HAPPEN, AND POSSIBLY HOW TO GO ABOUT THE RESEARCH FASTER. I also just signed up on Nairaland for this purpose.Soliciting contributions please.

PS: I Want to work with Python on this.
Re: Opinions On Speech Recognition Applying Machine Learning Algorithms by ogtavia(m): 10:34pm On May 11, 2018
where the programmers Nd IT egg heads at?
Re: Opinions On Speech Recognition Applying Machine Learning Algorithms by ChinenyeN(m): 7:00am On May 12, 2018
My own knowledge in the subject is limited, but I am currently actively pursuing self study to learn the tools necessary to achieve something like this. Machine learning can certainly be used to achieve speech recognition for Nigerian languages.

I cannot speak for certainty about the process, but here is what I understand about a possible route for using machine learning.

You record a large corpus of audio for each specific language. It should at least be large enough to cover the basics of communication in those languages. Once the audio corpus has been successfully collected, the audio data can be processed. This means transforming the audio into matrices representing the audio signal. The machine learning model can then be written and trained with that signal data.

Initially, you may want to use supervised learning to give the model a head start or a boost. In other words, help it to translate the audio signals to text, to prove that it can process and understand the signal data. Then, after the model is trained sufficiently to a satisfactory accuracy level, you can then attempt unsupervised learning to see if it can process a new audio sample that was not part of the corpus.

This is just a general, not very technical overview, because I am still developing my own knowledge and skills in this area, but I believe the process would be more or less along these lines.

Also, do note, you will have to do some research to determine the models that may be best suited for what you want to achieve. All machine learning models are not created equal.

Overall, what you want to achieve is certainly doable, but depending on your knowledge, it will take time.

1 Like 1 Share

Re: Opinions On Speech Recognition Applying Machine Learning Algorithms by 4dor: 8:23am On May 12, 2018
Natural Language Processing...Google it

1 Like

Re: Opinions On Speech Recognition Applying Machine Learning Algorithms by raymod170(m): 6:54pm On May 12, 2018
Speech recognition, speech synthesis, text to speech and natural languages procesors is all you need for your research.

Look up these and how to use them with the python language and frameworks as its doable.

use text to speech api it gives you the flexibility with other languages which may not be British or American English when pronounced by the user...but more close to native language pronouncation....study how speech Recogintion works on windows 7(it may be the easiest way to solve your problem)as well and implement that module with your knowledge of python


if you can't get an open source api for that purpose you can use your expertise to design a natural language processor which converts natural spoken words to the native language equivalent but mind you have to keep in mind that your api would have the whole native languages spoken in Nigeria for the communication to be effective... I may be wrong but you would do the same for text to speech, speech synthesis and speech recognition as well.... am guessing it's going to take year's to achieve... but what do I know..... if only you could ask one of Microsoft developers how long and how best they implemented their frameworks and api then you would have solved the problem by 10% going forward....

Am a .net person when it comes to full stack software development sorry but that's the best advice I can offer

best regards

1 Like

Re: Opinions On Speech Recognition Applying Machine Learning Algorithms by raymod170(m): 8:22pm On May 14, 2018
ChinenyeN:
My own knowledge in the subject is limited, but I am currently actively pursuing self study to learn the tools necessary to achieve something like this. Machine learning can certainly be used to achieve speech recognition for Nigerian languages.

I cannot speak for certainty about the process, but here is what I understand about a possible route for using machine learning.

You record a large corpus of audio for each specific language. It should at least be large enough to cover the basics of communication in those languages. Once the audio corpus has been successfully collected, the audio data can be processed. This means transforming the audio into matrices representing the audio signal. The machine learning model can then be written and trained with that signal data.

Initially, you may want to use supervised learning to give the model a head start or a boost. In other words, help it to translate the audio signals to text, to prove that it can process and understand the signal data. Then, after the model is trained sufficiently to a satisfactory accuracy level, you can then attempt unsupervised learning to see if it can process a new audio sample that was not part of the corpus.

This is just a general, not very technical overview, because I am still developing my own knowledge and skills in this area, but I believe the process would be more or less along these lines.

Also, do note, you will have to do some research to determine the models that may be best suited for what you want to achieve. All machine learning models are not created equal.

Overall, what you want to achieve is certainly doable, but depending on your knowledge, it will take time.

i dont understand you, are you saying he should record 520 languages spoken in Nigeria. how possible is that...

1 Like

(1) (Reply)

Real Python Contest / How Can I Become A Coding Expert Doing It By Myself? / App And Web Developer Needed For Employment

(Go Up)

Sections: politics (1) business autos (1) jobs (1) career education (1) romance computers phones travel sports fashion health
religion celebs tv-movies music-radio literature webmasters programming techmarket

Links: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Nairaland - Copyright © 2005 - 2024 Oluwaseun Osewa. All rights reserved. See How To Advertise. 23
Disclaimer: Every Nairaland member is solely responsible for anything that he/she posts or uploads on Nairaland.