: but it will not, and cannot, be the same as a natural recording.

The second does not imply the first. Compare programmatic instruments to ones that use sample banks in digital audio workstations.

In speech synthesis too you can take samples of sounds from a real person and process them to form new words and sentences. Of course it won't sound the same as if she acted out a whole line herself, and English is pretty complicated as languages go, but that doesn't mean it won't sound like her.

Look up 'Vocaloid' - it's a singing synth program and all the source material for the voice banks are from real people. Voice actors mostly.