mirror of
https://github.com/neodyland/sbv2-api.git
synced 2025-12-23 15:59:57 +00:00
Compare commits
22 Commits
commit-a66
...
commit-9f2
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
9f22694df0 | ||
|
|
62ba2c802f | ||
|
|
4f5b936f6f | ||
|
|
3c8efc716c | ||
|
|
e9ced32b70 | ||
|
|
e7a1575cbc | ||
|
|
873bbb77b6 | ||
|
|
1725863fca | ||
|
|
55f05580e4 | ||
|
|
320664eae2 | ||
|
|
87903827fa | ||
|
|
9b8e9dc39d | ||
|
|
bbc38081b6 | ||
|
|
0b822f704a | ||
|
|
132eb6386d | ||
|
|
ee56e9591d | ||
|
|
3194e599b2 | ||
|
|
00f4787f6e | ||
|
|
4b6c72aa51 | ||
|
|
7db6bb67a4 | ||
|
|
b3c75f973e | ||
|
|
e9529be559 |
2
.github/dependabot.yml
vendored
2
.github/dependabot.yml
vendored
@@ -9,7 +9,7 @@ updates:
|
||||
directory: "/" # Location of package manifests
|
||||
schedule:
|
||||
interval: "weekly"
|
||||
- package-ecosystem: "pnpm" # See documentation for possible values
|
||||
- package-ecosystem: "npm" # See documentation for possible values
|
||||
directory: "/" # Location of package manifests
|
||||
schedule:
|
||||
interval: "weekly"
|
||||
|
||||
32
Cargo.lock
generated
32
Cargo.lock
generated
@@ -995,7 +995,7 @@ dependencies = [
|
||||
"hyper",
|
||||
"libc",
|
||||
"pin-project-lite",
|
||||
"socket2",
|
||||
"socket2 0.5.9",
|
||||
"tokio",
|
||||
"tower-service",
|
||||
"tracing",
|
||||
@@ -1740,7 +1740,7 @@ checksum = "04744f49eae99ab78e0d5c0b603ab218f515ea8cfe5a456d7629ad883a3b6e7d"
|
||||
[[package]]
|
||||
name = "ort"
|
||||
version = "2.0.0-rc.10"
|
||||
source = "git+https://github.com/pykeio/ort.git#1e6f7ee1c8b056b00d280167ba172c96e78fcd1c"
|
||||
source = "git+https://github.com/pykeio/ort.git#f4ab181702495bff99a488322d3a8de0d7050349"
|
||||
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#1e6f7ee1c8b056b00d280167ba172c96e78fcd1c"
|
||||
source = "git+https://github.com/pykeio/ort.git#f4ab181702495bff99a488322d3a8de0d7050349"
|
||||
dependencies = [
|
||||
"flate2",
|
||||
"pkg-config",
|
||||
@@ -2430,9 +2430,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "serde_json"
|
||||
version = "1.0.140"
|
||||
version = "1.0.142"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "20068b6e96dc6c9bd23e01df8827e6c7e1f2fddd43c21810382803c136b99373"
|
||||
checksum = "030fedb782600dcbd6f02d479bf0d817ac3bb40d644745b769d6a96bc3afc5a7"
|
||||
dependencies = [
|
||||
"itoa",
|
||||
"memchr",
|
||||
@@ -2538,6 +2538,16 @@ dependencies = [
|
||||
"windows-sys 0.52.0",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "socket2"
|
||||
version = "0.6.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "233504af464074f9d066d7b5416c5f9b894a5862a6506e306f7b816cdd6f1807"
|
||||
dependencies = [
|
||||
"libc",
|
||||
"windows-sys 0.59.0",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "socks"
|
||||
version = "0.3.4"
|
||||
@@ -2756,9 +2766,9 @@ checksum = "1f3ccbac311fea05f86f61904b462b55fb3df8837a366dfc601a0161d0532f20"
|
||||
|
||||
[[package]]
|
||||
name = "tokenizers"
|
||||
version = "0.21.2"
|
||||
version = "0.21.4"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "4c3846d8588abed0daba25a0e47edd58ea15e450a6088b2575f5116fdb0b27ca"
|
||||
checksum = "a620b996116a59e184c2fa2dfd8251ea34a36d0a514758c6f966386bd2e03476"
|
||||
dependencies = [
|
||||
"ahash",
|
||||
"aho-corasick",
|
||||
@@ -2791,9 +2801,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "tokio"
|
||||
version = "1.46.1"
|
||||
version = "1.47.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "0cc3a2344dafbe23a245241fe8b09735b521110d30fcefbbd5feb1797ca35d17"
|
||||
checksum = "89e49afdadebb872d3145a5638b59eb0691ea23e46ca484037cfab3b76b95038"
|
||||
dependencies = [
|
||||
"backtrace",
|
||||
"bytes",
|
||||
@@ -2804,9 +2814,9 @@ dependencies = [
|
||||
"pin-project-lite",
|
||||
"signal-hook-registry",
|
||||
"slab",
|
||||
"socket2",
|
||||
"socket2 0.6.0",
|
||||
"tokio-macros",
|
||||
"windows-sys 0.52.0",
|
||||
"windows-sys 0.59.0",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
|
||||
@@ -16,7 +16,7 @@ env_logger.workspace = true
|
||||
log = "0.4.22"
|
||||
sbv2_core = { version = "0.2.0-alpha6", path = "../sbv2_core", features = ["aivmx"] }
|
||||
serde = { version = "1.0.210", features = ["derive"] }
|
||||
tokio = { version = "1.46.1", features = ["full"] }
|
||||
tokio = { version = "1.47.1", features = ["full"] }
|
||||
utoipa = { version = "5.4.0", features = ["axum_extras"] }
|
||||
utoipa-scalar = { version = "0.3.0", features = ["axum"] }
|
||||
|
||||
|
||||
@@ -22,10 +22,10 @@ once_cell.workspace = true
|
||||
ort = { git = "https://github.com/pykeio/ort.git", version = "2.0.0-rc.9", optional = true }
|
||||
regex = "1.10.6"
|
||||
serde = { version = "1.0.210", features = ["derive"] }
|
||||
serde_json = "1.0.128"
|
||||
serde_json = "1.0.142"
|
||||
tar = "0.4.41"
|
||||
thiserror = "2.0.11"
|
||||
tokenizers = { version = "0.21.2", default-features = false }
|
||||
tokenizers = { version = "0.21.4", default-features = false }
|
||||
zstd = "0.13.2"
|
||||
|
||||
[features]
|
||||
|
||||
@@ -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()))
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@@ -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
|
||||
|
||||
@@ -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()?;
|
||||
|
||||
Reference in New Issue
Block a user