gpt_sovits/text/
en.rs

1mod g2p_en;
2
3pub use g2p_en::*;
4use {
5    crate::{error::GSVError, text::get_phone_symbol},
6    log::debug,
7    std::borrow::Cow,
8};
9
10#[derive(PartialEq, Eq, Clone)]
11pub enum EnWord {
12    Word(String),
13    Punctuation(&'static str),
14}
15
16impl std::fmt::Debug for EnWord {
17    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
18        match self {
19            EnWord::Word(w) => write!(f, "\"{}\"", w),
20            EnWord::Punctuation(p) => write!(f, "\"{}\"", p),
21        }
22    }
23}
24
25#[derive(Debug, Default)]
26pub struct EnSentence {
27    pub phone_ids: Vec<i64>,
28    pub phones: Vec<Cow<'static, str>>,
29    pub word2ph: Vec<i32>,
30    pub text: Vec<EnWord>,
31}
32
33impl EnSentence {
34    pub fn g2p(&mut self, g2p_en: &mut G2pEn) -> Result<(), GSVError> {
35        self.phones.clear();
36        self.phone_ids.clear();
37        self.word2ph.clear();
38        for word in &self.text {
39            match word {
40                EnWord::Word(w) => {
41                    let phonemes = g2p_en.g2p(w)?;
42                    let mut cnt = 0;
43                    for ph in phonemes {
44                        self.phones.push(Cow::Owned(ph.clone()));
45                        self.phone_ids.push(get_phone_symbol(&ph));
46                        cnt += 1;
47                        if ph.contains("0")
48                            || ph.contains("1")
49                            || ph.contains("2")
50                            || ph.contains("3")
51                            || ph.contains("4")
52                        {
53                            self.word2ph.push(cnt);
54                            cnt = 0;
55                        }
56                    }
57                    if cnt > 0 {
58                        self.word2ph.push(cnt);
59                    }
60                }
61                EnWord::Punctuation(p) => {
62                    self.phones.push(Cow::Borrowed(p));
63                    self.phone_ids.push(get_phone_symbol(p));
64                    self.word2ph.push(1);
65                }
66            };
67        }
68        debug!("EnSentence phones: {:?}", self.phones);
69        debug!("EnSentence phone_ids: {:?}", self.phone_ids);
70        debug!("EnSentence word2ph: {:?}", self.word2ph);
71        Ok(())
72    }
73
74    pub fn build_phone(&self) -> Result<Vec<i64>, GSVError> {
75        Ok(self.phone_ids.clone())
76    }
77
78    pub fn get_text_string(&self) -> String {
79        let mut result = String::with_capacity(self.text.len() * 5); // Estimate capacity
80        for w in &self.text {
81            match w {
82                EnWord::Word(s) => result.push_str(s),
83                EnWord::Punctuation(p) => result.push_str(p),
84            }
85        }
86        result
87    }
88}
gpt_sovits/text/en.rs

gpt_sovits/text/
en.rs
