split the code for support voicevox

crates/sbv2_core/src/tts_util.rs

@@ -5,6 +5,19 @@ use crate::{jtalk, nlp, norm, tokenizer, utils};
 use hound::{SampleFormat, WavSpec, WavWriter};
 use ndarray::{concatenate, s, Array, Array1, Array2, Array3, Axis};
 use tokenizers::Tokenizer;
+
+pub fn preprocess_parse_text(
+    text: &str,
+    jtalk: &jtalk::JTalk,
+) -> Result<(Vec<String>, Vec<i32>, Vec<i32>)> {
+    let text = jtalk.num2word(text)?;
+    let normalized_text = norm::normalize_text(&text);
+
+    let process = jtalk.process_text(&normalized_text)?;
+    let result = process.g2p()?;
+    Ok(result)
+}
+
 /// Parse text and return the input for synthesize
 ///
 /// # Note
@@ -21,11 +34,7 @@ pub async fn parse_text(
         Box<dyn std::future::Future<Output = Result<ndarray::Array2<f32>>>>,
     >,
 ) -> Result<(Array2<f32>, Array1<i64>, Array1<i64>, Array1<i64>)> {
-    let text = jtalk.num2word(text)?;
-    let normalized_text = norm::normalize_text(&text);
-
-    let process = jtalk.process_text(&normalized_text)?;
-    let (phones, tones, mut word2ph) = process.g2p()?;
+    let (phones, tones, mut word2ph) = preprocess_parse_text(text, jtalk)?;
     let (phones, tones, lang_ids) = nlp::cleaned_text_to_sequence(phones, tones);
 
     let phones = utils::intersperse(&phones, 0);

crates/sbv2_voicevox/Cargo.toml

@@ -0,0 +1,14 @@
+[package]
+name = "sbv2_voicevox"
+version.workspace = true
+edition.workspace = true
+description.workspace = true
+license.workspace = true
+readme.workspace = true
+repository.workspace = true
+documentation.workspace = true
+
+[dependencies]
+anyhow.workspace = true
+axum = "0.8.1"
+sbv2_core = { version = "0.2.0-alpha6", path = "../sbv2_core" }

crates/sbv2_voicevox/README.md

@@ -0,0 +1,2 @@
+# sbv2-voicevox
+sbv2-apiをvoicevox化します。

crates/sbv2_voicevox/src/main.rs

@@ -0,0 +1,5 @@
+
+async fn main() -> anyhow::Result<()> {
+    println!("Hello, world!");
+    Ok(())
+}