Some suggestions for going a little bit further with this exercise, if you want to.
- Record your own speech and analyse that (instructions on recording speech can be found in the digit recogniser exercise.)
- record on different devices (your phone, your laptop, …) and compare the signals to see what differs – compare by listening, looking at the waveform, and inspecting the spectrogram. Can you figure out the cause of these differences?
- record close to the microphone, or far away (try it in a reverberant space, such as the bathroom!) – what changes? why?
- Analyse some of the synthetic speech signals in this zip file
- what are the differences between them? use different tools (your ears, the waveform, the spectrogram)
- in what ways are they similar or different to natural speech?
- could you tell they are synthetic just by looking at the spectrogram?