Compare commits

..

15 Commits

Author SHA1 Message Date
tuna2134
84e9118d99 Merge pull request #237 from neodyland/dependabot/cargo/ureq-3.1.0
build(deps): bump ureq from 3.0.12 to 3.1.0
2025-08-18 22:58:45 +09:00
tuna2134
3050cc1e99 Merge pull request #238 from neodyland/dependabot/cargo/thiserror-2.0.15
build(deps): bump thiserror from 2.0.12 to 2.0.15
2025-08-18 22:58:16 +09:00
tuna2134
d5fcacd799 Merge pull request #239 from neodyland/dependabot/cargo/ort-d269461
build(deps): bump ort from `f4ab181` to `d269461`
2025-08-18 22:58:01 +09:00
tuna2134
25ca89e341 Merge pull request #240 from neodyland/dependabot/cargo/anyhow-1.0.99
build(deps): bump anyhow from 1.0.98 to 1.0.99
2025-08-18 22:57:45 +09:00
dependabot[bot]
0c2a397775 build(deps): bump anyhow from 1.0.98 to 1.0.99
Bumps [anyhow](https://github.com/dtolnay/anyhow) from 1.0.98 to 1.0.99.
- [Release notes](https://github.com/dtolnay/anyhow/releases)
- [Commits](https://github.com/dtolnay/anyhow/compare/1.0.98...1.0.99)

---
updated-dependencies:
- dependency-name: anyhow
  dependency-version: 1.0.99
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
2025-08-18 11:08:26 +00:00
dependabot[bot]
470a0348fe build(deps): bump ort from f4ab181 to d269461
Bumps [ort](https://github.com/pykeio/ort) from `f4ab181` to `d269461`.
- [Release notes](https://github.com/pykeio/ort/releases)
- [Commits](f4ab181702...d269461e21)

---
updated-dependencies:
- dependency-name: ort
  dependency-version: d269461e2130b407589feff404025df25faeb3bb
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
2025-08-18 11:06:31 +00:00
dependabot[bot]
9a99b88b00 build(deps): bump thiserror from 2.0.12 to 2.0.15
Bumps [thiserror](https://github.com/dtolnay/thiserror) from 2.0.12 to 2.0.15.
- [Release notes](https://github.com/dtolnay/thiserror/releases)
- [Commits](https://github.com/dtolnay/thiserror/compare/2.0.12...2.0.15)

---
updated-dependencies:
- dependency-name: thiserror
  dependency-version: 2.0.15
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
2025-08-18 10:49:47 +00:00
dependabot[bot]
29f39f0795 build(deps): bump ureq from 3.0.12 to 3.1.0
Bumps [ureq](https://github.com/algesten/ureq) from 3.0.12 to 3.1.0.
- [Changelog](https://github.com/algesten/ureq/blob/main/CHANGELOG.md)
- [Commits](https://github.com/algesten/ureq/compare/3.0.12...3.1.0)

---
updated-dependencies:
- dependency-name: ureq
  dependency-version: 3.1.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
2025-08-18 10:49:32 +00:00
tuna2134
9f22694df0 Merge pull request #236 from neodyland/dependabot/cargo/ort-f4ab181 2025-08-12 07:48:08 +09:00
tuna2134
62ba2c802f Merge pull request #235 from kono-dada/fix/inplace-model-load 2025-08-11 23:46:42 +09:00
dependabot[bot]
4f5b936f6f build(deps): bump ort from 5f96a2d to f4ab181
Bumps [ort](https://github.com/pykeio/ort) from `5f96a2d` to `f4ab181`.
- [Release notes](https://github.com/pykeio/ort/releases)
- [Commits](5f96a2d585...f4ab181702)

---
updated-dependencies:
- dependency-name: ort
  dependency-version: f4ab181702495bff99a488322d3a8de0d7050349
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
2025-08-11 12:22:23 +00:00
kono-dada
3c8efc716c Fix: Load model in-place and safely evict sessions without removing entries
- Avoid removing and re-inserting model entries during load
- Preserve metadata (bytes, style_vectors) when evicting
- Ensure eviction targets a different loaded model, not always the first
- Reduce unnecessary memory allocations and keep list order stable
2025-08-11 16:31:57 +08:00
tuna2134
e9ced32b70 fix: streamline tone value handling in JTalkProcess 2025-08-11 17:30:46 +09:00
tuna2134
e7a1575cbc Merge pull request #233 from kono-dada/feature/stereo-output
feat: add stereo synthesis option via SBV2_FORCE_STEREO env var
2025-08-11 17:13:19 +09:00
kono-dada
873bbb77b6 feat: add stereo synthesis option via SBV2_FORCE_STEREO env var
Previously, synthesis output was fixed to mono (channels=1).
Now, setting the environment variable SBV2_FORCE_STEREO=1 forces stereo (2-channel) output.

This allows generating stereo audio without changing the code, useful for users needing dual-channel output.
2025-08-11 11:38:32 +08:00
6 changed files with 78 additions and 77 deletions

58
Cargo.lock generated
View File

@@ -92,9 +92,9 @@ dependencies = [
[[package]]
name = "anyhow"
version = "1.0.98"
version = "1.0.99"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e16d2d3311acee920a9eb8d33b8cbc1787ce4a264e85f964c2404b969bdcd487"
checksum = "b0674a1ddeecb70197781e945de4b3b8ffb61fa939a5597bcf48503737663100"
[[package]]
name = "atomic-waker"
@@ -1412,7 +1412,7 @@ dependencies = [
"reqwest",
"serde",
"tar",
"thiserror 2.0.12",
"thiserror 2.0.15",
"yada",
]
@@ -1740,7 +1740,7 @@ checksum = "04744f49eae99ab78e0d5c0b603ab218f515ea8cfe5a456d7629ad883a3b6e7d"
[[package]]
name = "ort"
version = "2.0.0-rc.10"
source = "git+https://github.com/pykeio/ort.git#5f96a2d5857c3fe9f06282dbf4bdcddbca6c5fe6"
source = "git+https://github.com/pykeio/ort.git#d269461e2130b407589feff404025df25faeb3bb"
dependencies = [
"libloading",
"ndarray",
@@ -1752,7 +1752,7 @@ dependencies = [
[[package]]
name = "ort-sys"
version = "2.0.0-rc.10"
source = "git+https://github.com/pykeio/ort.git#5f96a2d5857c3fe9f06282dbf4bdcddbca6c5fe6"
source = "git+https://github.com/pykeio/ort.git#d269461e2130b407589feff404025df25faeb3bb"
dependencies = [
"flate2",
"pkg-config",
@@ -1812,7 +1812,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "198db74531d58c70a361c42201efde7e2591e976d518caf7662a47dc5720e7b6"
dependencies = [
"memchr",
"thiserror 2.0.12",
"thiserror 2.0.15",
"ucd-trie",
]
@@ -2141,7 +2141,7 @@ checksum = "dd6f9d3d47bdd2ad6945c5015a226ec6155d0bcdfd8f7cd29f86b71f8de99d2b"
dependencies = [
"getrandom 0.2.16",
"libredox",
"thiserror 2.0.12",
"thiserror 2.0.15",
]
[[package]]
@@ -2352,7 +2352,7 @@ dependencies = [
"serde",
"serde_json",
"tar",
"thiserror 2.0.12",
"thiserror 2.0.15",
"tokenizers",
"ureq",
"zstd",
@@ -2710,11 +2710,11 @@ dependencies = [
[[package]]
name = "thiserror"
version = "2.0.12"
version = "2.0.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "567b8a2dae586314f7be2a752ec7474332959c6460e02bde30d702a66d488708"
checksum = "80d76d3f064b981389ecb4b6b7f45a0bf9fdac1d5b9204c7bd6714fecc302850"
dependencies = [
"thiserror-impl 2.0.12",
"thiserror-impl 2.0.15",
]
[[package]]
@@ -2730,9 +2730,9 @@ dependencies = [
[[package]]
name = "thiserror-impl"
version = "2.0.12"
version = "2.0.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7f7cf42b4507d8ea322120659672cf1b9dbb93f8f2d4ecfd6e51350ff5b17a1d"
checksum = "44d29feb33e986b6ea906bd9c3559a856983f92371b3eaa5e83782a351623de0"
dependencies = [
"proc-macro2",
"quote",
@@ -2793,7 +2793,7 @@ dependencies = [
"serde",
"serde_json",
"spm_precompiled",
"thiserror 2.0.12",
"thiserror 2.0.15",
"unicode-normalization-alignments",
"unicode-segmentation",
"unicode_categories",
@@ -2997,9 +2997,9 @@ checksum = "8ecb6da28b8a351d773b68d5825ac39017e680750f980f3a1a85cd8dd28a47c1"
[[package]]
name = "ureq"
version = "3.0.12"
version = "3.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9f0fde9bc91026e381155f8c67cb354bcd35260b2f4a29bcc84639f762760c39"
checksum = "00432f493971db5d8e47a65aeb3b02f8226b9b11f1450ff86bb772776ebadd70"
dependencies = [
"base64 0.22.1",
"der",
@@ -3013,15 +3013,15 @@ dependencies = [
"socks",
"ureq-proto",
"utf-8",
"webpki-root-certs 0.26.11",
"webpki-roots 0.26.11",
"webpki-root-certs",
"webpki-roots",
]
[[package]]
name = "ureq-proto"
version = "0.4.2"
version = "0.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "59db78ad1923f2b1be62b6da81fe80b173605ca0d57f85da2e005382adf693f7"
checksum = "c5b6cabebbecc4c45189ab06b52f956206cea7d8c8a20851c35a85cb169224cc"
dependencies = [
"base64 0.22.1",
"http",
@@ -3221,15 +3221,6 @@ dependencies = [
"wasm-bindgen",
]
[[package]]
name = "webpki-root-certs"
version = "0.26.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "75c7f0ef91146ebfb530314f5f1d24528d7f0767efbfd31dce919275413e393e"
dependencies = [
"webpki-root-certs 1.0.0",
]
[[package]]
name = "webpki-root-certs"
version = "1.0.0"
@@ -3239,15 +3230,6 @@ dependencies = [
"rustls-pki-types",
]
[[package]]
name = "webpki-roots"
version = "0.26.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "521bc38abb08001b01866da9f51eb7c5d647a19260e00054a8c7fd5f9e57f7a9"
dependencies = [
"webpki-roots 1.0.0",
]
[[package]]
name = "webpki-roots"
version = "1.0.0"

View File

@@ -12,7 +12,7 @@ repository = "https://github.com/neodyland/sbv2-api"
documentation = "https://docs.rs/sbv2_core"
[workspace.dependencies]
anyhow = "1.0.96"
anyhow = "1.0.99"
dotenvy = "0.15.7"
env_logger = "0.11.6"
ndarray = "0.16.1"

View File

@@ -24,7 +24,7 @@ regex = "1.10.6"
serde = { version = "1.0.210", features = ["derive"] }
serde_json = "1.0.142"
tar = "0.4.41"
thiserror = "2.0.11"
thiserror = "2.0.15"
tokenizers = { version = "0.21.4", default-features = false }
zstd = "0.13.2"
@@ -44,4 +44,4 @@ base64 = ["dep:base64"]
[build-dependencies]
dirs = "6.0.0"
ureq = "3.0.12"
ureq = "3.1.0"

View File

@@ -127,20 +127,20 @@ impl JTalkProcess {
Ok(phone_tone_list)
} else if tone_values.len() == 2 {
if tone_values == hash_set![0, 1] {
return Ok(phone_tone_list);
Ok(phone_tone_list)
} else if tone_values == hash_set![-1, 0] {
return Ok(phone_tone_list
Ok(phone_tone_list
.iter()
.map(|x| {
let new_tone = if x.1 == -1 { 0 } else { 1 };
(x.0.clone(), new_tone)
})
.collect());
.collect())
} else {
return Err(Error::ValueError("Invalid tone values 0".to_string()));
Err(Error::ValueError("Invalid tone values 0".to_string()))
}
} else {
return Err(Error::ValueError("Invalid tone values 1".to_string()));
Err(Error::ValueError("Invalid tone values 1".to_string()))
}
}

View File

@@ -240,39 +240,43 @@ impl TTSModelHolder {
}
fn find_and_load_model<I: Into<TTSIdent>>(&mut self, ident: I) -> Result<bool> {
let ident = ident.into();
let (bytes, style_vectors) = {
let model = self
.models
.iter()
.find(|m| m.ident == ident)
.ok_or(Error::ModelNotFoundError(ident.to_string()))?;
if model.vits2.is_some() {
return Ok(true);
}
(model.bytes.clone().unwrap(), model.style_vectors.clone())
};
self.unload(ident.clone());
let s = model::load_model(&bytes, false)?;
if let Some(max) = self.max_loaded_models {
if self.models.iter().filter(|x| x.vits2.is_some()).count() >= max {
self.unload(self.models.first().unwrap().ident.clone());
}
}
self.models.push(TTSModel {
bytes: Some(bytes.to_vec()),
vits2: Some(s),
style_vectors,
ident: ident.clone(),
});
let model = self
// Locate target model entry
let target_index = self
.models
.iter()
.find(|m| m.ident == ident)
.position(|m| m.ident == ident)
.ok_or(Error::ModelNotFoundError(ident.to_string()))?;
if model.vits2.is_some() {
// Already loaded
if self.models[target_index].vits2.is_some() {
return Ok(true);
}
Err(Error::ModelNotFoundError(ident.to_string()))
// Get bytes to build a Session
let bytes = self.models[target_index]
.bytes
.clone()
.ok_or(Error::ModelNotFoundError(ident.to_string()))?;
// Enforce max loaded models by evicting a different loaded model's session, not removing the entry
if let Some(max) = self.max_loaded_models {
let loaded_count = self.models.iter().filter(|m| m.vits2.is_some()).count();
if loaded_count >= max {
if let Some(evict_index) = self
.models
.iter()
.position(|m| m.vits2.is_some() && m.ident != ident)
{
// Drop only the session to free memory; keep bytes/style for future reload
self.models[evict_index].vits2 = None;
}
}
}
// Build and set session in-place for the target model
let s = model::load_model(&bytes, false)?;
self.models[target_index].vits2 = Some(s);
Ok(true)
}
/// Get style vector by style id and weight

View File

@@ -173,8 +173,15 @@ pub fn parse_text_blocking(
}
pub fn array_to_vec(audio_array: Array3<f32>) -> Result<Vec<u8>> {
// If SBV2_FORCE_STEREO is set ("1"/"true"), duplicate mono to stereo
let force_stereo = std::env::var("SBV2_FORCE_STEREO")
.ok()
.map(|v| matches!(v.as_str(), "1" | "true" | "TRUE" | "True"))
.unwrap_or(false);
let channels: u16 = if force_stereo { 2 } else { 1 };
let spec = WavSpec {
channels: 1,
channels,
sample_rate: 44100,
bits_per_sample: 32,
sample_format: SampleFormat::Float,
@@ -183,8 +190,16 @@ pub fn array_to_vec(audio_array: Array3<f32>) -> Result<Vec<u8>> {
let mut writer = WavWriter::new(&mut cursor, spec)?;
for i in 0..audio_array.shape()[0] {
let output = audio_array.slice(s![i, 0, ..]).to_vec();
for sample in output {
writer.write_sample(sample)?;
if force_stereo {
for sample in output {
// Write to Left and Right channels
writer.write_sample(sample)?;
writer.write_sample(sample)?;
}
} else {
for sample in output {
writer.write_sample(sample)?;
}
}
}
writer.finalize()?;