Empowering the largest knowledge workforce on Earth.
The success of modern speech recognition software is based largely on the availability of massive corpuses of recorded voice. Languages and dialects which have not been made a part of such corpuses to date are of interest to researchers and scientists for two reasons: 1) Interactive Voice Response (IVR) systems for new languages require such samples to be implemented and 2) out-of-sample speech samples are needed to verify new speech recognition algorithms (even if such algorithms are targeted to eventually be applied to existing speech corpuses like English).txteagle users speak a wide range of languages, many of which have not been recorded into modern speech corpuses yet.
Client: Nokia
Target Worker: Kenyan Luo Speaker
Task: Voice Recording
Question: [Luo] Please say “give me directions” after the tone.