- This topic has 1 reply, 2 voices, and was last updated 8 years, 7 months ago by .
Viewing 1 reply thread
Viewing 1 reply thread
- You must be logged in to reply to this topic.
› Forums › Speech Synthesis › Unit selection › Target in unit selection
According to Hunt(1996), target cost is an estimate of the difference between a database unit u, and the target t. I understand the unit u is a candidate unit in the database, but what exactly is the target? Shouldn’t it be the same as the unit we select, which resulting in a target cost of 0?
For instance, if the target is phoneme /n/ with an F0 of 121 and a duration of 60, shouldn’t we just select a unit that is a phoneme /n/ with an F0 of 121 and a duration of 60? Is it because that the features for targets are continuous values instead of discrete values, so that it is hard to find perfect mapping between unit and target? Or is it because there are other features that need to be taken into consideration? Or I misunderstand anything?
I think you’ve just missed one simple point: it will not be possible, in general, to find any candidates in the database that have exactly the same linguistic specification as the target.
In your example, where you are using phone-sized units an an ASF target cost, your target specification is “phoneme /n/ with an F0 of 121Hz and a duration of 60ms”. It is very unlikely that we will find a candidate with exactly those values. Imagine that we find these candidates:
None of these will have zero target cost.
Some forums are only available if you are logged in. Searching will only return results from those forums if you log in.
Copyright © 2024 · Balance Child Theme on Genesis Framework · WordPress · Log in