<< Chapter < Page Chapter >> Page >
Listed in this module are output results and analysis of test-audio samples from various speakers, fed to a trained neural network.

Results

The following are some example outputs from the neural network from various test speakers. The output displaysrelative strengths of different types of accents prevalent in a particular subject. All test inputs were not used in the trainingmatrix. Overall, approximately 20 tests were conducted with about an 80% success rate. Those that failed tended to with good reason(either inadequate recording quality, or speakers who did not provide accurate information about what their accent is comprisedof – a common issue with subjects who have lived in multiple places).

The charts below show accents in the following order: Northern US, Texan US, Russian, Farsi, and Mandarin

Test 1: chinese subject

Chinese subject

Chinese Subject (accent order: Northern US, Texan US, Russian, Farsi, and Mandarin)

Here our network has successfully picked out the accent of our subject. Secondarily, the network picked up on a slight Texan accent, possibly showing the influence of location on the subject (The sample was recorded in Texas).

Test 2: iranian subject

Iranian subject

Iranian Subject (accent order: Northern US, Texan US, Russian, Farsi, and Mandarin)

Again our network has successfully picked out the accent of our subject. Once again, this sample was recorded in Texas, which could account for the secondary influence of a Texan accent in the subject.

Test 3: chinese subject

Chinese subject

Chinese Subject (accent order: Northern US, Texan US, Russian, Farsi, and Mandarin)

Once again, the network successfully picks up on the subjects primary accent as well as influence of a Texan accent (this sample was also recorded in Texas).

Test 4: chinese subject

Chinese subject

Chinese Subject (accent order: Northern US, Texan US, Russian, Farsi, and Mandarin)

A successful test showing little or no influence from other accents in the network.

Test 5: american subject (hybrid of regions)

American subject (hybrid)

American Subject - Hybrid (accent order: Northern US, Texan US, Russian, Farsi, and Mandarin)

Results from a subject who has lived all over -- mainly in Texas, who's accent appears to sound more Northern (which seems relatively true if one listens to the source recording).

Test 6: russian subject

Russian subject

Russian Subject (accent order: Northern US, Texan US, Russian, Farsi, and Mandarin)

Successful test of a Russian subject with strong influences of a Northern US accent.

Test 7: russian subject

Russian subject

Russian Subject (accent order: Northern US, Texan US, Russian, Farsi, and Mandarin)

Another successful test of a Russian subject with strong influences of a Northern US accent.

Test 8: cantonese subject

Cantonese subject

Cantonese Subject (accent order: Northern US, Texan US, Russian, Farsi, and Mandarin)

Successful region-based test of a Cantonese subject who has been living in the US.

Test 1: korean subject

Korean subject

Korean Subject (accent order: Northern US, Texan US, Russian, Farsi, and Mandarin)

An interesting example of throwing an accent at the network that doesn't fit into any of the categories.

Get Jobilize Job Search Mobile App in your pocket Now!

Get it on Google Play Download on the App Store Now




Source:  OpenStax, Accent classification using neural networks. OpenStax CNX. Dec 15, 2005 Download for free at http://cnx.org/content/col10320/1.1
Google Play and the Google Play logo are trademarks of Google Inc.

Notification Switch

Would you like to follow the 'Accent classification using neural networks' conversation and receive update notifications?

Ask