Voice modulator (the Bowtie of Detective Conan)

[Med Jacer]

I'm seeking help to study some deeply the subject of voice identification, voice imitating, And if possible creating the bowtie of Conan or some something related to it.

How to create a system (program or algorithm) that identifies a person by his voice relying on its unique caracteristics (tone..) ?
Is it possible to imitate this voice ?

Any ideas or hints are welcome

 

[jedishrfu]

Here's a wiki article that can get you started in understanding the field:

https://en.wikipedia.org/wiki/Speaker_recognition

We did a student project last year where Apache Spark and some Deep Learning tools were used to construct a gunshot recognition network which would have similar characteristics to your voice recognition project.

So Apache Spark and DeepLearning4Java would be some software you could look at. There are other tools based on Pyrhon or Matlab that may also be of interest. You can find more information by researching Deep Learning or Machine Learning topics on Google and Wikipedia.

https://en.wikipedia.org/wiki/Deep_learning

https://en.wikipedia.org/wiki/Machine_learning

https://en.wikipedia.org/wiki/Apache_Spark

https://en.wikipedia.org/wiki/Deeplearning4j

There are also videos on Youtube by 3blue1brown on how neural nets work that also may be of interest.

https://www.google.com/search?newwi...i131i20i264k1j0i20i264k1j0i10k1.0.CmMSmTvsb8M