new: cleaned_sequence

sbv2_core/src/jtalk.rs

@@ -128,7 +128,6 @@ impl JTalkProcess {
 
         let mut phone_tone_list =
             JTalkProcess::align_tones(phone_w_punct, phone_tone_list_wo_punct)?;
-        println!("{:?}", phone_tone_list);
 
         let mut sep_tokenized: Vec<Vec<String>> = Vec::new();
         for i in 0..seq_text.len() {

sbv2_core/src/lib.rs

@@ -1,8 +1,9 @@
 pub mod bert;
 pub mod error;
+pub mod jtalk;
 pub mod mora;
+pub mod nlp;
 pub mod norm;
-pub mod text;
 
 pub fn add(left: usize, right: usize) -> usize {
     left + right

sbv2_core/src/main.rs

@@ -1,19 +1,19 @@
-use sbv2_core::{bert, error, text};
+use sbv2_core::{bert, error, jtalk};
 
 fn main() -> error::Result<()> {
     let text = "こんにちは,世界!";
 
-    let normalized_text = text::normalize_text(text);
+    let normalized_text = jtalk::normalize_text(text);
     println!("{}", normalized_text);
 
-    let jtalk = text::JTalk::new()?;
+    let jtalk = jtalk::JTalk::new()?;
     let (phones, tones, _) = jtalk.g2p(&normalized_text)?;
     println!("{:?}", tones);
 
-    let tokenizer = text::get_tokenizer()?;
+    let tokenizer = jtalk::get_tokenizer()?;
     println!("{:?}", tokenizer);
 
-    let (token_ids, attention_masks) = text::tokenize(&normalized_text, &tokenizer)?;
+    let (token_ids, attention_masks) = jtalk::tokenize(&normalized_text, &tokenizer)?;
     println!("{:?}", token_ids);
 
     let session = bert::load_model()?;

sbv2_core/src/nlp.rs

@@ -0,0 +1,24 @@
+use crate::norm::SYMBOLS;
+use once_cell::sync::Lazy;
+use std::collections::HashMap;
+
+static SYMBOL_TO_ID: Lazy<HashMap<String, i32>> = Lazy::new(|| {
+    let mut map = HashMap::new();
+    for (i, symbols) in SYMBOLS.iter().enumerate() {
+        map.insert(symbols.to_string(), i as i32);
+    }
+    map
+});
+
+pub fn cleaned_text_to_sequence(
+    cleaned_phones: Vec<String>,
+    tones: Vec<i32>,
+) -> (Vec<i32>, Vec<i32>, Vec<i32>) {
+    let phones: Vec<i32> = cleaned_phones
+        .iter()
+        .map(|phone| SYMBOL_TO_ID.get(phone).unwrap())
+        .collect();
+    let tones: Vec<i32> = tones.iter().map(|tone| tone + 6).collect();
+    let lang_ids: Vec<i32> = vec![1; phones.len()];
+    (phones, tones, lang_ids)
+}

sbv2_core/src/norm.rs

@@ -69,7 +69,26 @@ __PUNCTUATION_CLEANUP_PATTERN = re.compile(
 )
 */
 
+pub const JP_SYMBOLS: [&str; 42] = [
+    "N", "a", "a:", "b", "by", "ch", "d", "dy", "e", "e:", "f", "g", "gy", "h", "hy", "i", "i:",
+    "j", "k", "ky", "m", "my", "n", "ny", "o", "o:", "p", "py", "q", "r", "ry", "s", "sh", "t",
+    "ts", "ty", "u", "u:", "w", "y", "z", "zy",
+];
+
 pub static PUNCTUATIONS: [&str; 7] = ["!", "?", "…", ",", ".", "'", "-"];
+pub static PUNCTUATION_SYMBOLS: Lazy<Vec<&str>> = Lazy::new(|| {
+    let mut symbols = PUNCTUATIONS.to_vec();
+    symbols.append(&mut vec!["SP", "UNK"]);
+    symbols
+});
+const PAD: &str = "_";
+pub static SYMBOLS: Lazy<Vec<&str>> = Lazy::new(|| {
+    let mut symbols = JP_SYMBOLS.to_vec();
+    symbols.append(&mut JP_SYMBOLS.to_vec());
+    symbols.append(&mut PUNCTUATION_SYMBOLS.to_vec());
+    symbols
+});
+
 static PUNCTUATION_CLEANUP_PATTERN: Lazy<regex::Regex> = Lazy::new(|| {
     let pattern = r"[^\u{3040}-\u{309F}\u{30A0}-\u{30FF}\u{4E00}-\u{9FFF}\u{3400}-\u{4DBF}\u{3005}"
         .to_owned()