A. Build ASR based on Kaldi, trained using Librispeech and Mozilla libraries, and a video library recalled from the web.
B. Be able to cancel out background noise, such as music, to accurately assess just the speech element.
C. KPIs such as pronounciation (consonants and vowels), intonation, speed must be tracked, per time and content.
D. A custom word list based on user performance must be auto generated.
GUI /dashboard must
E. Graphically display the KPIs for increased user usability and navigation ease.
F. Content, including online videos, to be recalled into a user dashboard
G. Loaded content, mostly videos, must be able to play while being capable of the on/offing of receiving and assessing user voice input, synced to loaded content.
Answer the questions below to prove that you have read the bid.
1. To start the project, need your advice on developing the deliverable for accessibility over multiple platforms. Your suggestion(s)?
2. Do you prefer using Kaldi, Sphinx, or another solution? Why?
3. Privacy is very important to our market, because of regulations. We are refraining from Google speech recognition for example because we work in a regulated industry. What security measures will you implement to ensure a secure system?
4. Explain what WER is.
5. Please summarize this project. Explain to me in a sentence what we are trying to build.
Hello, I read the description of your project thoroughly.
I understand your requirements basically, and I have experiences of similar project.
I am professional Website builder and Mobile App developer, and talented Angular/Nodejs/Vuejs especially.
Also I can provide other windows desktop application using .NET techniques.
You can see my abilities in my recent reviews, on my profile page.
https://www.freelancer.com/u/luiswilliam
Now, I have enough time to work for your project and I'm ready to go with you.
Usually, I would provide long term support on the work I did.
Looking forward your contacting.
Thanks!
1540286023660
Hello,
I am pleasure with your job for ASR based on Kaldi.
Thank you for the job posting. It’s a pleasure to meet you.
I’d really like to work with you on this one if possible!
I do have a couple of questions, but first I’d like to make you an offer and some background so you can check my work out.
I have been developing kind of project within 10+ years so I’m fluent experience to handle project.
You’ll get all the expected stuff like a great professional service and a fast turnaround, at a bit less, and I get a bit more exposure.
If the above offer sounds like something you would be interested in, I’d love to hear from you.
Best regards,
Georgy
1. To start the project, need your advice on developing the deliverable for accessibility over multiple platforms. Your suggestion(s)?
For multiple platform deliverable i believe we have to develop it in java so cross platform compatibility can achieved
2. Do you prefer using Kaldi, Sphinx, or another solution? Why?
I will prefer Sphinnx or Kaldi
3. Privacy is very important to our market, because of regulations. We are refraining from Google speech recognition for example because we work in a regulated industry. What security measures will you implement to ensure a secure system?
We are using the library our own and all code will deleted after delivery to you
4. Explain what WER is.
Word Error Rate it's about how much accuracy it have
5. Please summarize this project. Explain to me in a sentence what we are trying to build.
As i understand it's all about the project it's all about voice reorganization and voice to text .
- the ASR can be implemented as a web application with Python in the backend, allowing it to run on any platform.
- I use Python libraries such as Deepspeech which is opensource and not solutions as Kladi or Sphinx
- security will depend on how the user interact with the application, having it as web application means that the code will be secured in the back end and only the input sound will be sent to it.
- WER is the Word Error Rate which is similar to Levenshtein distance or edit distance , when we compare the text generated by the ASR to the actual text we count the number of substitutions, deletions, and insertions over the number of words.
- we are building an automatic speech recognition model to have speech as input and outputs text.
I already worked on training an ASR using Librispeech data, and i have a good experience in this area. I worked on multilingual speech recognition model not only English.
there are many things to discuss for such a project, hopefully we can do that over the messages.
I think the project needs to be more an hourly rate project, because it has many parts, and many tasks to be done.
Welcome to my profile.
I'm rich experienced with the developments of Desktop & Mobile App and Website .
Your satisfaction is a duty for me, so I'll do our best on your project.
? 8+ years of experience in developing desktop applications for the PC
• C/C++ Programming
• C#/VB.NET & JAVA Programming
? 6+ years of experience in developing Mobile Apps
• Native Platform(All version of Android/iOS)
• Hybrid / React Native
• Backend API & 3rd Party API Manipulation. (Google, Instagram, Paypal and etc)
? 5+ years of experience in developing Games
• Unity3D, Cocos2d
? 5+ years of experience in developing Websites
• PHP / JavaScript(Angular/Node)
• Wordpress, Joomla, Prestashop ( Plugins, Theme)
• CSS Frameworks ( Bootstrap , Foundation )
• Database(Mysql/SQLite/Oracle/MongoDB)
? Why I am?
* Honest and Polite
* Daily / Weekly result, Responsible communication anytime.
* 6 months Free bug fixing and maintenance after development.
Thanks