NETtalk (artificial neural network)

The intent behind NETtalk was to construct simplified models that might shed light on the complexity of learning human level cognitive tasks, and their implementation as a connectionist model that could also learn to perform a comparable task.

The output of the network was a stream of phonemes, which fed into DECtalk to produce audible speech, It achieved popular success, appearing on the Today show.

[3]: 115 The training dataset was a 20,008-word subset of the Brown Corpus, with manually annotated phoneme and stress for each letter.

[4][5] After it was run successfully on this, the authors tried it on a phonological transcription of an interview with a young Latino boy from a barrio in Los Angeles.

[3]: 115 The original NETtalk was implemented on a Ridge 32, which took 0.275 seconds per learning step (one forward and one backward pass).

21 units encode for articulatory features (point of articulation, voicing, vowel height, etc.)

The output of the network degrades, but remains understandable, when some hidden neurons are removed.

[9] NETtalk was created to explore the mechanisms of learning to correctly pronounce English text.

NETtalk does not specifically model the image processing stages and letter recognition of the visual cortex.