Representing Speech Data

• Speech signal has a lot of variation

• Hard to recognise speech from the raw signal

• Need to reduce the signal somehow

• Various transforms are available

• Many based on FFT