Re: [Espeak-general] Improvment for Cantonese

[Jonathan D. <jo...@js...>]

On 26 Aug, <hgn...@ya...> wrote:

> I've made some significant improvement to Cantonese.

> I have annotated phonetic symbols for about 40,000
> words with multi-sound characters and generated a new
> zhy_list.

> And I have also modified some vowels.

Thanks.

1. There is a problem with the multi-character sequences.
The phonetic translation must be a single string without spaces.  Words
are separated by ||

For example:
  h'ou||j'in4||s'oeng3||hyun3 

or, without word breaks:
  h'ou|j'in4|s'oeng3|hyun3 

not
  h'ou j'in4 s'oeng3 hyun3 

Is it possible for you to create the  zhy_list  file using || or |
instead of spaces?


2. In some sequences, for example:
  h'ou||j'in4||s'oeng3||hyun3 

A vowel does not have a stress marker (eg. "hyun3"), so the vowel is
unstressed.  Is this intentional?


3. In the voice file,  espeak-data/voices/zhy
Does:
  words 1

sound better than:
  words 5