モーフィング機能を追加する

Cargo.toml

@@ -10,6 +10,7 @@ anyhow = "1.0.65"
 assert_cmd = "2.0.8"
 async-std = "1.12.0"
 async_zip = "0.0.11"
+az = "1.2.1"
 binstall-tar = "0.4.39"
 bytes = "1.1.0"
 cbindgen = "0.24.3"
@@ -46,6 +47,7 @@ log = "0.4.17"
 nanoid = "0.4.0"
 ndarray = "0.15.6"
 ndarray-stats = "0.5.1"
+num-traits = "0.2.15"
 octocrab = { version = "0.19.0", default-features = false }
 once_cell = "1.18.0"
 ouroboros = "0.18.0"
@@ -94,6 +96,10 @@ rev = "a16714ce16dec76fd0e3041a7acfa484921db3b5"
 git = "https://github.com/VOICEVOX/process_path.git"
 rev = "de226a26e8e18edbdb1d6f986afe37bbbf35fbf4"
 
+[workspace.dependencies.world]
+git = "https://github.com/White-Green/WORLD_rs.git"
+rev = "2337a30bfa47eebd32ef418c60ae5c7b39e43b99"
+
 [workspace.package]
 version = "0.0.0"
 edition = "2021"

crates/voicevox_core/Cargo.toml

@@ -11,6 +11,7 @@ directml = ["onnxruntime/directml"]
 [dependencies]
 anyhow.workspace = true
 async_zip = { workspace = true, features = ["full"] }
+az.workspace = true
 derive-getters.workspace = true
 derive-new.workspace = true
 derive_more.workspace = true
@@ -24,6 +25,7 @@ indexmap = { workspace = true, features = ["serde"] }
 itertools.workspace = true
 nanoid.workspace = true
 ndarray.workspace = true
+num-traits.workspace = true
 once_cell.workspace = true
 onnxruntime.workspace = true
 open_jtalk.workspace = true
@@ -38,6 +40,7 @@ tokio = { workspace = true, features = ["rt"] } # FIXME: feature-gateする
 tracing.workspace = true
 uuid = { workspace = true, features = ["v4", "serde"] }
 voicevox_core_macros = { path = "../voicevox_core_macros" }
+world.workspace = true
 zip.workspace = true
 
 [dev-dependencies]

crates/voicevox_core/src/engine/mod.rs

@@ -3,11 +3,13 @@ mod full_context_label;
 mod kana_parser;
 mod model;
 mod mora_list;
+mod morph;
 pub(crate) mod open_jtalk;
 
 pub(crate) use self::acoustic_feature_extractor::OjtPhoneme;
 pub(crate) use self::full_context_label::{FullContextLabelError, Utterance};
 pub(crate) use self::kana_parser::{create_kana, parse_kana, KanaParseError};
-pub use self::model::{AccentPhraseModel, AudioQueryModel, MoraModel};
+pub use self::model::{AccentPhraseModel, AudioQueryModel, MoraModel, MorphableTargetInfo};
 pub(crate) use self::mora_list::mora2text;
+pub(crate) use self::morph::{MorphError, MorphingPair};
 pub use self::open_jtalk::FullcontextExtractor;

crates/voicevox_core/src/engine/model.rs

@@ -82,6 +82,11 @@ impl AudioQueryModel {
     }
 }
 
+#[derive(Deserialize, Serialize)]
+pub struct MorphableTargetInfo {
+    pub is_morphable: bool,
+}
+
 #[cfg(test)]
 mod tests {
     use pretty_assertions::assert_eq;