[Anon Student]

For human speech production, do we talk about it from a phonetic and biological standpoint and for tts, do we describe the source filter model?

[Anon Student]

Allophones are the representation of phonemes in actual speech sounds. A phoneme can make up different allophones. Is this correct?

[Anon Student]

My mistake they are used. Phonemes are units that distinguish sounds and are used in g2p that makes use of LTS rules to determine pronunciation of words. Is this correct?

[Anon Student]

No, we use diphones instead as they capture co-articulation. The place and manner of articulation as well as whether the source is voiced or not will affect the acoustic properties of a speech sound.

Is this knowledge used in TTS through the concatenation of diphones that capture co-articulation?

[Anon Student]

Is the question basically asking how different phones are differentiated in human speech and in speech synthesis?

[Anon Student]

I had a followup question to answering ‘What in the speech signal differentiates different phones?’. Having looked at the modules and noted how the different phones are differentiated, what angle should we take in using this question to explain how concatenative TTS differs from human speech production and why this might lead to errors? I’m struggling to see a connection point between the two.

[Anon Student]

For the background section, are we trying to explain the festival background based within each of the headings in the marking scheme or include the content of the headings in a overall explanation?

[Anon Student]

Thank you, I also wanted to double check, do words in our diagrams contribute to the word count?

[Author]

Posts