fix

話者指定を追加

.github/workflows/CI.yml

@@ -140,9 +140,6 @@ jobs:
     strategy:
       matrix:
         tag: [cpu, cuda]
-        platform:
-        - linux/amd64
-        - linux/arm64
     steps:
       - uses: actions/checkout@v4
       - name: Set up QEMU
@@ -163,4 +160,4 @@ jobs:
           tags: |
             ghcr.io/${{ github.repository }}:${{ matrix.tag }}
           file: docker/${{ matrix.tag }}.Dockerfile
-          platforms: ${{ matrix.platform }}
+          platforms: linux/amd64, linux/arm64

Cargo.lock

@@ -4,9 +4,9 @@ version = 3
 
 [[package]]
 name = "addr2line"
-version = "0.24.1"
+version = "0.24.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f5fb1d8e4442bd405fdfd1dacb42792696b0cf9cb15882e5d097b742a676d375"
+checksum = "dfbe277e56a376000877090da837660b4427aad530e3028d44e0bffe4f89a1c1"
 dependencies = [
  "gimli",
 ]
@@ -77,15 +77,15 @@ dependencies = [
 
 [[package]]
 name = "anyhow"
-version = "1.0.90"
+version = "1.0.93"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "37bf3594c4c988a53154954629820791dde498571819ae4ca50ca811e060cc95"
+checksum = "4c95c10ba0b00a02636238b814946408b1322d5ac4760326e6fb8ec956d85775"
 
 [[package]]
 name = "async-trait"
-version = "0.1.82"
+version = "0.1.83"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a27b8a3a6e1a44fa4c8baf1f653e4172e81486d4941f2237e20dc2d0cf4ddff1"
+checksum = "721cae7de5c34fbb2acd27e21e6d2cf7b886dce0c27388d46c4e6c47ea4318dd"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -94,9 +94,9 @@ dependencies = [
 
 [[package]]
 name = "autocfg"
-version = "1.3.0"
+version = "1.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0c4b4d0bd25bd0b74681c0ad21497610ce1b7c91b1022cd21c80c6fbdd9476b0"
+checksum = "ace50bade8e6234aa140d9a2f552bbee1db4d353f69b8217bc503490fc1a9f26"
 
 [[package]]
 name = "axum"
@@ -126,7 +126,7 @@ dependencies = [
  "serde_urlencoded",
  "sync_wrapper 1.0.1",
  "tokio",
- "tower 0.5.1",
+ "tower",
  "tower-layer",
  "tower-service",
  "tracing",
@@ -239,15 +239,15 @@ checksum = "1fd0f2584146f6f2ef48085050886acf353beff7305ebd1ae69500e27c67f64b"
 
 [[package]]
 name = "bytes"
-version = "1.7.1"
+version = "1.8.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8318a53db07bb3f8dca91a600466bdb3f2eaadeedfdbcf02e1accbad9271ba50"
+checksum = "9ac0150caa2ae65ca5bd83f25c7de183dea78d4d366469f148435e2acfbad0da"
 
 [[package]]
 name = "cc"
-version = "1.1.18"
+version = "1.1.31"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b62ac837cdb5cb22e10a256099b4fc502b1dfe560cb282963a974d7abd80e476"
+checksum = "c2e7962b54006dcfcc61cb72735f4d89bb97061dd6a7ed882ec6b8ee53714c6f"
 dependencies = [
  "jobserver",
  "libc",
@@ -396,18 +396,18 @@ dependencies = [
 
 [[package]]
 name = "derive_builder"
-version = "0.20.1"
+version = "0.20.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cd33f37ee6a119146a1781d3356a7c26028f83d779b2e04ecd45fdc75c76877b"
+checksum = "507dfb09ea8b7fa618fcf76e953f4f5e192547945816d5358edffe39f6f94947"
 dependencies = [
  "derive_builder_macro",
 ]
 
 [[package]]
 name = "derive_builder_core"
-version = "0.20.1"
+version = "0.20.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7431fa049613920234f22c47fdc33e6cf3ee83067091ea4277a3f8c4587aae38"
+checksum = "2d5bcf7b024d6835cfb3d473887cd966994907effbe9227e8c8219824d06c4e8"
 dependencies = [
  "darling",
  "proc-macro2",
@@ -417,9 +417,9 @@ dependencies = [
 
 [[package]]
 name = "derive_builder_macro"
-version = "0.20.1"
+version = "0.20.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4abae7035bf79b9877b779505d8cf3749285b80c43941eda66604841889451dc"
+checksum = "ab63b0e2bf4d5928aff72e83a7dace85d7bba5fe12dcc3c5a572d78caffd3f3c"
 dependencies = [
  "derive_builder_core",
  "syn",
@@ -608,9 +608,9 @@ dependencies = [
 
 [[package]]
 name = "flate2"
-version = "1.0.33"
+version = "1.0.34"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "324a1be68054ef05ad64b861cc9eaf1d623d2d8cb25b4bf2cb9cdd902b4bf253"
+checksum = "a1b589b4dc103969ad3cf85c950899926ec64300a1a46d76c03a6072957036f0"
 dependencies = [
  "crc32fast",
  "miniz_oxide",
@@ -633,30 +633,30 @@ dependencies = [
 
 [[package]]
 name = "futures-channel"
-version = "0.3.30"
+version = "0.3.31"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "eac8f7d7865dcb88bd4373ab671c8cf4508703796caa2b1985a9ca867b3fcb78"
+checksum = "2dff15bf788c671c1934e366d07e30c1814a8ef514e1af724a602e8a2fbe1b10"
 dependencies = [
  "futures-core",
 ]
 
 [[package]]
 name = "futures-core"
-version = "0.3.30"
+version = "0.3.31"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dfc6580bb841c5a68e9ef15c77ccc837b40a7504914d52e47b8b0e9bbda25a1d"
+checksum = "05f29059c0c2090612e8d742178b0580d2dc940c837851ad723096f87af6663e"
 
 [[package]]
 name = "futures-task"
-version = "0.3.30"
+version = "0.3.31"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "38d84fa142264698cdce1a9f9172cf383a0c82de1bddcf3092901442c4097004"
+checksum = "f90f7dce0722e95104fcb095585910c0977252f286e354b5e3bd38902cd99988"
 
 [[package]]
 name = "futures-util"
-version = "0.3.30"
+version = "0.3.31"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3d6401deb83407ab3da39eba7e33987a73c3df0c82b4bb5813ee871c19c41d48"
+checksum = "9fa08315bb612088cc391249efdc3bc77536f16c91f6cf495e6fbe85b20a4a81"
 dependencies = [
  "futures-core",
  "futures-task",
@@ -689,9 +689,9 @@ dependencies = [
 
 [[package]]
 name = "gimli"
-version = "0.31.0"
+version = "0.31.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "32085ea23f3234fc7846555e85283ba4de91e21016dc0455a16286d87a292d64"
+checksum = "07e28edb80900c19c28f1072f2e8aeca7fa06b23cd4169cefe1af5aa3260783f"
 
 [[package]]
 name = "glob"
@@ -711,9 +711,9 @@ dependencies = [
 
 [[package]]
 name = "hashbrown"
-version = "0.14.5"
+version = "0.15.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e5274423e17b7c9fc20b6e7e208532f9b19825d82dfd615708b70edd83df41f1"
+checksum = "1e087f84d4f86bf4b218b927129862374b72199ae7d8657835f1e89000eea4fb"
 
 [[package]]
 name = "heck"
@@ -769,9 +769,9 @@ dependencies = [
 
 [[package]]
 name = "httparse"
-version = "1.9.4"
+version = "1.9.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0fcc0b4a115bf80b728eb8ea024ad5bd707b615bfed49e0665b6e0f86fd082d9"
+checksum = "7d71d3574edd2771538b901e6549113b4006ece66150fb69c0fb6d9a2adae946"
 
 [[package]]
 name = "httpdate"
@@ -787,9 +787,9 @@ checksum = "9a3a5bfb195931eeb336b2a7b4d761daec841b97f947d34394601737a7bba5e4"
 
 [[package]]
 name = "hyper"
-version = "1.4.1"
+version = "1.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "50dfd22e0e76d0f662d429a5f80fcaf3855009297eab6a0a9f8543834744ba05"
+checksum = "bbbff0a806a4728c99295b254c8838933b5b082d75e3cb70c8dab21fdfbcfa9a"
 dependencies = [
  "bytes",
  "futures-channel",
@@ -806,9 +806,9 @@ dependencies = [
 
 [[package]]
 name = "hyper-util"
-version = "0.1.8"
+version = "0.1.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "da62f120a8a37763efb0cf8fdf264b884c7b8b9ac8660b900c8661030c00e6ba"
+checksum = "41296eb09f183ac68eec06e03cdbea2e759633d4067b2f6552fc2e009bcad08b"
 dependencies = [
  "bytes",
  "futures-util",
@@ -817,7 +817,6 @@ dependencies = [
  "hyper",
  "pin-project-lite",
  "tokio",
- "tower 0.4.13",
  "tower-service",
 ]
 
@@ -839,9 +838,9 @@ dependencies = [
 
 [[package]]
 name = "indexmap"
-version = "2.5.0"
+version = "2.6.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "68b900aa2f7301e21c36462b170ee99994de34dff39a4a6a528e80e7376d07e5"
+checksum = "707907fe3c25f5424cce2cb7e1cbcafee6bdbe735ca90ef77c29e84591e5b9da"
 dependencies = [
  "equivalent",
  "hashbrown",
@@ -1059,9 +1058,9 @@ checksum = "bbd2bcb4c963f2ddae06a2efc7e9f3591312473c50c6685e1f298068316e66fe"
 
 [[package]]
 name = "libc"
-version = "0.2.158"
+version = "0.2.161"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d8adc4bb1803a324070e64a98ae98f38934d91957a99cfb3a43dcbc01bc56439"
+checksum = "8e9489c2807c139ffd9c1794f4af0ebe86a828db53ecdc7fea2111d0fed085d1"
 
 [[package]]
 name = "libloading"
@@ -1420,9 +1419,9 @@ checksum = "830b246a0e5f20af87141b25c173cd1b609bd7779a4617d6ec582abaf90870f3"
 
 [[package]]
 name = "object"
-version = "0.36.4"
+version = "0.36.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "084f1a5821ac4c651660a94a7153d27ac9d8a53736203f58b31945ded098070a"
+checksum = "aedf0a2d09c573ed1d8d85b30c119153926a2b36dce0ab28322c09a117a4683e"
 dependencies = [
  "memchr",
 ]
@@ -1457,8 +1456,8 @@ dependencies = [
 
 [[package]]
 name = "ort"
-version = "2.0.0-rc.6"
+version = "2.0.0-rc.8"
-source = "git+https://github.com/pykeio/ort.git#7ad5b84e5a42cb6399635d95d88fc81aa38fddf3"
+source = "git+https://github.com/pykeio/ort.git#585efda15393c949010903990bd0a53e7d3d6ffe"
 dependencies = [
  "half",
  "libloading",
@@ -1469,8 +1468,8 @@ dependencies = [
 
 [[package]]
 name = "ort-sys"
-version = "2.0.0-rc.6"
+version = "2.0.0-rc.8"
-source = "git+https://github.com/pykeio/ort.git#7ad5b84e5a42cb6399635d95d88fc81aa38fddf3"
+source = "git+https://github.com/pykeio/ort.git#585efda15393c949010903990bd0a53e7d3d6ffe"
 dependencies = [
  "flate2",
  "pkg-config",
@@ -1556,26 +1555,6 @@ dependencies = [
  "siphasher",
 ]
 
-[[package]]
-name = "pin-project"
-version = "1.1.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b6bf43b791c5b9e34c3d182969b4abb522f9343702850a2e57f460d00d09b4b3"
-dependencies = [
- "pin-project-internal",
-]
-
-[[package]]
-name = "pin-project-internal"
-version = "1.1.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2f38a4412a78282e09a2cf38d195ea5420d15ba0602cb375210efbc877243965"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn",
-]
-
 [[package]]
 name = "pin-project-lite"
 version = "0.2.14"
@@ -1590,21 +1569,21 @@ checksum = "8b870d8c151b6f2fb93e84a13146138f05d02ed11c7e7c54f8826aaaf7c9f184"
 
 [[package]]
 name = "pkg-config"
-version = "0.3.30"
+version = "0.3.31"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d231b230927b5e4ad203db57bbcbee2802f6bce620b1e4a9024a07d94e2907ec"
+checksum = "953ec861398dccce10c670dfeaf3ec4911ca479e9c02154b3a215178c5f566f2"
 
 [[package]]
 name = "portable-atomic"
-version = "1.7.0"
+version = "1.9.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "da544ee218f0d287a911e9c99a39a8c9bc8fcad3cb8db5959940044ecfc67265"
+checksum = "cc9c68a3f6da06753e9335d63e27f6b9754dd1920d941135b7ea8224f141adb2"
 
 [[package]]
 name = "portable-atomic-util"
-version = "0.2.2"
+version = "0.2.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fcdd8420072e66d54a407b3316991fe946ce3ab1083a7f575b2463866624704d"
+checksum = "90a7d5beecc52a491b54d6dd05c7a45ba1801666a5baad9fdbfc6fef8d2d206c"
 dependencies = [
  "portable-atomic",
 ]
@@ -1620,18 +1599,18 @@ dependencies = [
 
 [[package]]
 name = "proc-macro2"
-version = "1.0.86"
+version = "1.0.89"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5e719e8df665df0d1c8fbfd238015744736151d4445ec0836b8e628aae103b77"
+checksum = "f139b0662de085916d1fb67d2b4169d1addddda1919e696f3252b740b629986e"
 dependencies = [
  "unicode-ident",
 ]
 
 [[package]]
 name = "pyo3"
-version = "0.22.5"
+version = "0.22.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3d922163ba1f79c04bc49073ba7b32fd5a8d3b76a87c955921234b8e77333c51"
+checksum = "f402062616ab18202ae8319da13fa4279883a2b8a9d9f83f20dbade813ce1884"
 dependencies = [
  "anyhow",
  "cfg-if",
@@ -1648,9 +1627,9 @@ dependencies = [
 
 [[package]]
 name = "pyo3-build-config"
-version = "0.22.5"
+version = "0.22.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "bc38c5feeb496c8321091edf3d63e9a6829eab4b863b4a6a65f26f3e9cc6b179"
+checksum = "b14b5775b5ff446dd1056212d778012cbe8a0fbffd368029fd9e25b514479c38"
 dependencies = [
  "once_cell",
  "target-lexicon",
@@ -1658,9 +1637,9 @@ dependencies = [
 
 [[package]]
 name = "pyo3-ffi"
-version = "0.22.5"
+version = "0.22.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "94845622d88ae274d2729fcefc850e63d7a3ddff5e3ce11bd88486db9f1d357d"
+checksum = "9ab5bcf04a2cdcbb50c7d6105de943f543f9ed92af55818fd17b660390fc8636"
 dependencies = [
  "libc",
  "pyo3-build-config",
@@ -1668,9 +1647,9 @@ dependencies = [
 
 [[package]]
 name = "pyo3-macros"
-version = "0.22.5"
+version = "0.22.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e655aad15e09b94ffdb3ce3d217acf652e26bbc37697ef012f5e5e348c716e5e"
+checksum = "0fd24d897903a9e6d80b968368a34e1525aeb719d568dba8b3d4bfa5dc67d453"
 dependencies = [
  "proc-macro2",
  "pyo3-macros-backend",
@@ -1680,9 +1659,9 @@ dependencies = [
 
 [[package]]
 name = "pyo3-macros-backend"
-version = "0.22.5"
+version = "0.22.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ae1e3f09eecd94618f60a455a23def79f79eba4dc561a97324bf9ac8c6df30ce"
+checksum = "36c011a03ba1e50152b4b394b479826cad97e7a21eb52df179cd91ac411cbfbe"
 dependencies = [
  "heck",
  "proc-macro2",
@@ -1769,18 +1748,18 @@ dependencies = [
 
 [[package]]
 name = "redox_syscall"
-version = "0.5.3"
+version = "0.5.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2a908a6e00f1fdd0dfd9c0eb08ce85126f6d8bbda50017e74bc4a4b7d4a926a4"
+checksum = "9b6dfecf2c74bce2466cabf93f6664d6998a69eb21e39f4207930065b27b771f"
 dependencies = [
  "bitflags 2.6.0",
 ]
 
 [[package]]
 name = "regex"
-version = "1.11.0"
+version = "1.11.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "38200e5ee88914975b69f657f0801b6f6dccafd44fd9326302a4aaeecfacb1d8"
+checksum = "b544ef1b4eac5dc2db33ea63606ae9ffcfac26c1416a2806ae0bf5f56b201191"
 dependencies = [
  "aho-corasick",
  "memchr",
@@ -1828,9 +1807,9 @@ checksum = "719b953e2095829ee67db738b3bfa9fa368c94900df327b3f07fe6e794d2fe1f"
 
 [[package]]
 name = "rustix"
-version = "0.38.36"
+version = "0.38.37"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3f55e80d50763938498dd5ebb18647174e0c76dc38c5505294bb224624f30f36"
+checksum = "8acb788b847c24f28525660c4d7758620a7210875711f79e7f663cc152726811"
 dependencies = [
  "bitflags 2.6.0",
  "errno",
@@ -1841,9 +1820,9 @@ dependencies = [
 
 [[package]]
 name = "rustls"
-version = "0.23.13"
+version = "0.23.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f2dabaac7466917e566adb06783a81ca48944c6898a1b08b9374106dd671f4c8"
+checksum = "5fbb44d7acc4e873d613422379f69f237a1b141928c02f6bc6ccfddddc2d7993"
 dependencies = [
  "log",
  "once_cell",
@@ -1856,9 +1835,9 @@ dependencies = [
 
 [[package]]
 name = "rustls-pki-types"
-version = "1.8.0"
+version = "1.10.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fc0a2ce646f8655401bb81e7927b812614bd5d91dbc968696be50603510fcaf0"
+checksum = "16f1201b3c9a7ee8039bcadc17b7e605e2945b27eee7631788c1bd2b0643674b"
 
 [[package]]
 name = "rustls-webpki"
@@ -1873,9 +1852,9 @@ dependencies = [
 
 [[package]]
 name = "rustversion"
-version = "1.0.17"
+version = "1.0.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "955d28af4278de8121b7ebeb796b6a45735dc01436d898801014aced2773a3d6"
+checksum = "0e819f2bc632f285be6d7cd36e25940d45b2391dd6d9b939e79de557f7014248"
 
 [[package]]
 name = "ryu"
@@ -1951,18 +1930,18 @@ checksum = "94143f37725109f92c262ed2cf5e59bce7498c01bcc1502d7b9afe439a4e9f49"
 
 [[package]]
 name = "serde"
-version = "1.0.210"
+version = "1.0.215"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c8e3592472072e6e22e0a54d5904d9febf8508f65fb8552499a1abc7d1078c3a"
+checksum = "6513c1ad0b11a9376da888e3e0baa0077f1aed55c17f50e7b2397136129fb88f"
 dependencies = [
  "serde_derive",
 ]
 
 [[package]]
 name = "serde_derive"
-version = "1.0.210"
+version = "1.0.215"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "243902eda00fad750862fc144cea25caca5e20d615af0a81bee94ca738f1df1f"
+checksum = "ad1e866f866923f252f05c889987993144fb74e722403468a4ebd70c3cd756c0"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -2116,9 +2095,9 @@ checksum = "13c2bddecc57b384dee18652358fb23172facb8a2c51ccc10d74c157bdea3292"
 
 [[package]]
 name = "syn"
-version = "2.0.77"
+version = "2.0.87"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9f35bcdf61fd8e7be6caf75f429fdca8beb3ed76584befb503b1569faee373ed"
+checksum = "25aa4ce346d03a6dcd68dd8b4010bcb74e54e62c90c573f394c46eae99aba32d"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -2139,9 +2118,9 @@ checksum = "a7065abeca94b6a8a577f9bd45aa0867a2238b74e8eb67cf10d492bc39351394"
 
 [[package]]
 name = "tar"
-version = "0.4.42"
+version = "0.4.43"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4ff6c40d3aedb5e06b57c6f669ad17ab063dd1e63d977c6a88e7f4dfa4f04020"
+checksum = "c65998313f8e17d0d553d28f91a0df93e4dbbbf770279c7bc21ca0f09ea1a1f6"
 dependencies = [
  "filetime",
  "libc",
@@ -2156,18 +2135,18 @@ checksum = "61c41af27dd6d1e27b1b16b489db798443478cef1f06a660c96db617ba5de3b1"
 
 [[package]]
 name = "thiserror"
-version = "1.0.64"
+version = "1.0.69"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d50af8abc119fb8bb6dbabcfa89656f46f84aa0ac7688088608076ad2b459a84"
+checksum = "b6aaf5339b578ea85b50e080feb250a3e8ae8cfcdff9a461c9ec2904bc923f52"
 dependencies = [
  "thiserror-impl",
 ]
 
 [[package]]
 name = "thiserror-impl"
-version = "1.0.64"
+version = "1.0.69"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "08904e7672f5eb876eaaf87e0ce17857500934f4981c4a0ab2b4aa98baac7fc3"
+checksum = "4fee6c4efc90059e10f81e6d42c60a18f76588c3d74cb83a0b242a2b6c7504c1"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -2191,9 +2170,9 @@ checksum = "1f3ccbac311fea05f86f61904b462b55fb3df8837a366dfc601a0161d0532f20"
 
 [[package]]
 name = "tokenizers"
-version = "0.20.1"
+version = "0.20.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b172ffa9a2e5c31bbddc940cd5725d933ced983a9333bbebc4c7eda3bbce1557"
+checksum = "67b67c92f6d705e2a1d106fb0b28c696f9074901a9c656ee5d9f5de204c39bf7"
 dependencies = [
  "aho-corasick",
  "derive_builder",
@@ -2224,9 +2203,9 @@ dependencies = [
 
 [[package]]
 name = "tokio"
-version = "1.40.0"
+version = "1.41.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e2b070231665d27ad9ec9b8df639893f46727666c6767db40317fbe920a5d998"
+checksum = "22cfb5bee7a6a52939ca9224d6ac897bb669134078daa8735560897f69de4d33"
 dependencies = [
  "backtrace",
  "bytes",
@@ -2251,21 +2230,6 @@ dependencies = [
  "syn",
 ]
 
-[[package]]
-name = "tower"
-version = "0.4.13"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b8fa9be0de6cf49e536ce1851f987bd21a43b771b09473c3549a6c853db37c1c"
-dependencies = [
- "futures-core",
- "futures-util",
- "pin-project",
- "pin-project-lite",
- "tokio",
- "tower-layer",
- "tower-service",
-]
-
 [[package]]
 name = "tower"
 version = "0.5.1"
@@ -2322,9 +2286,9 @@ checksum = "42ff0bf0c66b8238c6f3b578df37d0b7848e55df8577b3f74f92a69acceeb825"
 
 [[package]]
 name = "unicode-bidi"
-version = "0.3.15"
+version = "0.3.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "08f95100a766bf4f8f28f90d77e0a5461bbdb219042e7679bebe79004fed8d75"
+checksum = "5ab17db44d7388991a428b2ee655ce0c212e862eff1768a455c58f9aad6e7893"
 
 [[package]]
 name = "unicode-ident"
@@ -2334,9 +2298,9 @@ checksum = "e91b56cd4cadaeb79bbf1a5645f6b4f8dc5bde8834ad5894a8db35fda9efa1fe"
 
 [[package]]
 name = "unicode-normalization"
-version = "0.1.23"
+version = "0.1.24"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a56d1686db2308d901306f92a263857ef59ea39678a5458e7cb17f01415101f5"
+checksum = "5033c97c4262335cded6d6fc3e5c18ab755e1a3dc96376350f3d8e9f009ad956"
 dependencies = [
  "tinyvec",
 ]
@@ -2352,15 +2316,15 @@ dependencies = [
 
 [[package]]
 name = "unicode-segmentation"
-version = "1.11.0"
+version = "1.12.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d4c87d22b6e3f4a18d4d40ef354e97c90fcb14dd91d7dc0aa9d8a1172ebf7202"
+checksum = "f6ccf251212114b54433ec949fd6a7841275f9ada20dddd2f29e9ceea4501493"
 
 [[package]]
 name = "unicode-width"
-version = "0.1.13"
+version = "0.1.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0336d538f7abc86d282a4189614dfaa90810dfc2c6f6427eaf88e16311dd225d"
+checksum = "7dd6e30e90baa6f72411720665d41d89b9a3d039dc45b8faea1ddd07f617f6af"
 
 [[package]]
 name = "unicode_categories"
@@ -2415,9 +2379,9 @@ checksum = "06abde3611657adf66d383f00b093d7faecc7fa57071cce2578660c9f1010821"
 
 [[package]]
 name = "utoipa"
-version = "5.1.1"
+version = "5.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8861811f7213bb866cd02319acb69a15b0ef8ca46874e805bd92d488c779036a"
+checksum = "514a48569e4e21c86d0b84b5612b5e73c0b2cf09db63260134ba426d4e8ea714"
 dependencies = [
  "indexmap",
  "serde",
@@ -2427,9 +2391,9 @@ dependencies = [
 
 [[package]]
 name = "utoipa-gen"
-version = "5.1.1"
+version = "5.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5fadf94f07d67df4b15e6490dd9a9d59d7374849413e7f137eafe52fdcbd0db5"
+checksum = "5629efe65599d0ccd5d493688cbf6e03aa7c1da07fe59ff97cf5977ed0637f66"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -2530,9 +2494,9 @@ checksum = "65fc09f10666a9f147042251e0dda9c18f166ff7de300607007e96bdebc1068d"
 
 [[package]]
 name = "web-sys"
-version = "0.3.70"
+version = "0.3.72"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "26fdeaafd9bd129f65e7c031593c24d62186301e0c72c8978fa1678be7d532c0"
+checksum = "f6488b90108c040df0fe62fa815cbdee25124641df01814dd7282749234c6112"
 dependencies = [
  "js-sys",
  "wasm-bindgen",
@@ -2540,9 +2504,9 @@ dependencies = [
 
 [[package]]
 name = "webpki-roots"
-version = "0.26.5"
+version = "0.26.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0bd24728e5af82c6c4ec1b66ac4844bdf8156257fccda846ec58b42cd0cdbe6a"
+checksum = "841c67bff177718f1d4dfefde8d8f0e78f9b6589319ba88312f567fc5841a958"
 dependencies = [
  "rustls-pki-types",
 ]

sbv2_api/src/main.rs

@@ -40,6 +40,14 @@ fn length_default() -> f32 {
     1.0
 }
 
+fn style_id_default() -> i32 {
+    0
+}
+
+fn speaker_id_default() -> i64 {
+    0
+}
+
 #[derive(Deserialize, ToSchema)]
 struct SynthesizeRequest {
     text: String,
@@ -48,6 +56,10 @@ struct SynthesizeRequest {
     sdp_ratio: f32,
     #[serde(default = "length_default")]
     length_scale: f32,
+    #[serde(default = "style_id_default")]
+    style_id: i32,
+    #[serde(default = "speaker_id_default")]
+    speaker_id: i64,
 }
 
 #[utoipa::path(
@@ -65,15 +77,18 @@ async fn synthesize(
         ident,
         sdp_ratio,
         length_scale,
+        style_id,
+        speaker_id,
     }): Json<SynthesizeRequest>,
 ) -> AppResult<impl IntoResponse> {
     log::debug!("processing request: text={text}, ident={ident}, sdp_ratio={sdp_ratio}, length_scale={length_scale}");
     let buffer = {
-        let tts_model = state.tts_model.lock().await;
+        let mut tts_model = state.tts_model.lock().await;
         tts_model.easy_synthesize(
             &ident,
             &text,
-            0,
+            style_id,
+            speaker_id,
             SynthesizeOptions {
                 sdp_ratio,
                 length_scale,
@@ -94,6 +109,9 @@ impl AppState {
         let mut tts_model = TTSModelHolder::new(
             &fs::read(env::var("BERT_MODEL_PATH")?).await?,
             &fs::read(env::var("TOKENIZER_PATH")?).await?,
+            env::var("HOLDER_MAX_LOADED_MODElS")
+                .ok()
+                .and_then(|x| x.parse().ok()),
         )?;
         let models = env::var("MODELS_PATH").unwrap_or("models".to_string());
         let mut f = fs::read_dir(&models).await?;

sbv2_bindings/src/sbv2.rs

@@ -23,10 +23,15 @@ pub struct TTSModel {
 
 #[pymethods]
 impl TTSModel {
+    #[pyo3(signature = (bert_model_bytes, tokenizer_bytes, max_loaded_models=None))]
     #[new]
-    fn new(bert_model_bytes: Vec<u8>, tokenizer_bytes: Vec<u8>) -> anyhow::Result<Self> {
+    fn new(
+        bert_model_bytes: Vec<u8>,
+        tokenizer_bytes: Vec<u8>,
+        max_loaded_models: Option<usize>,
+    ) -> anyhow::Result<Self> {
         Ok(Self {
-            model: TTSModelHolder::new(bert_model_bytes, tokenizer_bytes)?,
+            model: TTSModelHolder::new(bert_model_bytes, tokenizer_bytes, max_loaded_models)?,
         })
     }
 
@@ -38,10 +43,21 @@ impl TTSModel {
     ///     BERTモデルのパス
     /// tokenizer_path : str
     ///     トークナイザーのパス
+    /// max_loaded_models: int | None
+    ///     同時にVRAMに存在するモデルの数
+    #[pyo3(signature = (bert_model_path, tokenizer_path, max_loaded_models=None))]
     #[staticmethod]
-    fn from_path(bert_model_path: String, tokenizer_path: String) -> anyhow::Result<Self> {
+    fn from_path(
+        bert_model_path: String,
+        tokenizer_path: String,
+        max_loaded_models: Option<usize>,
+    ) -> anyhow::Result<Self> {
         Ok(Self {
-            model: TTSModelHolder::new(fs::read(bert_model_path)?, fs::read(tokenizer_path)?)?,
+            model: TTSModelHolder::new(
+                fs::read(bert_model_path)?,
+                fs::read(tokenizer_path)?,
+                max_loaded_models,
+            )?,
         })
     }
 
@@ -121,11 +137,12 @@ impl TTSModel {
     /// voice_data : bytes
     ///     音声データ
     fn synthesize<'p>(
-        &'p self,
+        &'p mut self,
         py: Python<'p>,
         text: String,
         ident: String,
         style_id: i32,
+        speaker_id: i64,
         sdp_ratio: f32,
         length_scale: f32,
     ) -> anyhow::Result<Bound<PyBytes>> {
@@ -133,6 +150,7 @@ impl TTSModel {
             ident.as_str(),
             &text,
             style_id,
+            speaker_id,
             SynthesizeOptions {
                 sdp_ratio,
                 length_scale,

sbv2_core/Cargo.toml

@@ -17,7 +17,7 @@ jpreprocess = { version = "0.10.0", features = ["naist-jdic"] }
 ndarray.workspace = true
 num_cpus = "1.16.0"
 once_cell.workspace = true
-ort = { git = "https://github.com/pykeio/ort.git", version = "2.0.0-rc.6", optional = true }
+ort = { git = "https://github.com/pykeio/ort.git", version = "2.0.0-rc.8", optional = true }
 regex = "1.10.6"
 serde = { version = "1.0.210", features = ["derive"] }
 serde_json = "1.0.128"

sbv2_core/src/main.rs

@@ -11,10 +11,14 @@ fn main_inner() -> anyhow::Result<()> {
     let mut tts_holder = tts::TTSModelHolder::new(
         &fs::read(env::var("BERT_MODEL_PATH")?)?,
         &fs::read(env::var("TOKENIZER_PATH")?)?,
+        env::var("HOLDER_MAX_LOADED_MODElS")
+            .ok()
+            .and_then(|x| x.parse().ok()),
     )?;
     tts_holder.load_sbv2file(ident, fs::read(env::var("MODEL_PATH")?)?)?;
 
-    let audio = tts_holder.easy_synthesize(ident, &text, 0, tts::SynthesizeOptions::default())?;
+    let audio =
+        tts_holder.easy_synthesize(ident, &text, 0, 0, tts::SynthesizeOptions::default())?;
     fs::write("output.wav", audio)?;
 
     Ok(())

sbv2_core/src/model.rs

@@ -52,6 +52,7 @@ pub fn synthesize(
     session: &Session,
     bert_ori: Array2<f32>,
     x_tst: Array1<i64>,
+    sid: Array1<i64>,
     tones: Array1<i64>,
     lang_ids: Array1<i64>,
     style_vector: Array1<f32>,
@@ -67,7 +68,7 @@ pub fn synthesize(
     let outputs = session.run(ort::inputs! {
         "x_tst" => x_tst,
         "x_tst_lengths" => x_tst_lengths,
-        "sid" => array![0_i64],
+        "sid" => sid,
         "tones" => tones,
         "language" => lang_ids,
         "bert" => bert,

sbv2_core/src/tts.rs

@@ -24,9 +24,10 @@ where
 }
 
 pub struct TTSModel {
-    vits2: Session,
+    vits2: Option<Session>,
     style_vectors: Array2<f32>,
     ident: TTSIdent,
+    bytes: Option<Vec<u8>>,
 }
 
 /// High-level Style-Bert-VITS2's API
@@ -35,6 +36,7 @@ pub struct TTSModelHolder {
     bert: Session,
     models: Vec<TTSModel>,
     jtalk: jtalk::JTalk,
+    max_loaded_models: Option<usize>,
 }
 
 impl TTSModelHolder {
@@ -43,9 +45,13 @@ impl TTSModelHolder {
     /// # Examples
     ///
     /// ```rs
-    /// let mut tts_holder = TTSModelHolder::new(std::fs::read("deberta.onnx")?, std::fs::read("tokenizer.json")?)?;
+    /// let mut tts_holder = TTSModelHolder::new(std::fs::read("deberta.onnx")?, std::fs::read("tokenizer.json")?, None)?;
     /// ```
-    pub fn new<P: AsRef<[u8]>>(bert_model_bytes: P, tokenizer_bytes: P) -> Result<Self> {
+    pub fn new<P: AsRef<[u8]>>(
+        bert_model_bytes: P,
+        tokenizer_bytes: P,
+        max_loaded_models: Option<usize>,
+    ) -> Result<Self> {
         let bert = model::load_model(bert_model_bytes, true)?;
         let jtalk = jtalk::JTalk::new()?;
         let tokenizer = tokenizer::get_tokenizer(tokenizer_bytes)?;
@@ -54,6 +60,7 @@ impl TTSModelHolder {
             models: vec![],
             jtalk,
             tokenizer,
+            max_loaded_models,
         })
     }
 
@@ -94,10 +101,25 @@ impl TTSModelHolder {
     ) -> Result<()> {
         let ident = ident.into();
         if self.find_model(ident.clone()).is_err() {
+            let mut load = true;
+            if let Some(max) = self.max_loaded_models {
+                if self.models.iter().filter(|x| x.vits2.is_some()).count() >= max {
+                    load = false;
+                }
+            }
             self.models.push(TTSModel {
-                vits2: model::load_model(vits2_bytes, false)?,
+                vits2: if load {
+                    Some(model::load_model(&vits2_bytes, false)?)
+                } else {
+                    None
+                },
                 style_vectors: style::load_style(style_vectors_bytes)?,
                 ident,
+                bytes: if self.max_loaded_models.is_some() {
+                    Some(vits2_bytes.as_ref().to_vec())
+                } else {
+                    None
+                },
             })
         }
         Ok(())
@@ -145,6 +167,42 @@ impl TTSModelHolder {
             .find(|m| m.ident == ident)
             .ok_or(Error::ModelNotFoundError(ident.to_string()))
     }
+    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
+            .models
+            .iter()
+            .find(|m| m.ident == ident)
+            .ok_or(Error::ModelNotFoundError(ident.to_string()))?;
+        if model.vits2.is_some() {
+            return Ok(true);
+        }
+        Err(Error::ModelNotFoundError(ident.to_string()))
+    }
 
     /// Get style vector by style id and weight
     ///
@@ -167,12 +225,19 @@ impl TTSModelHolder {
     /// let audio = tts_holder.easy_synthesize("tsukuyomi", "こんにちは", 0, SynthesizeOptions::default())?;
     /// ```
     pub fn easy_synthesize<I: Into<TTSIdent> + Copy>(
-        &self,
+        &mut self,
         ident: I,
         text: &str,
         style_id: i32,
+        speaker_id: i64,
         options: SynthesizeOptions,
     ) -> Result<Vec<u8>> {
+        self.find_and_load_model(ident)?;
+        let vits2 = &self
+            .find_model(ident)?
+            .vits2
+            .as_ref()
+            .ok_or(Error::ModelNotFoundError(ident.into().to_string()))?;
         let style_vector = self.get_style_vector(ident, style_id, options.style_weight)?;
         let audio_array = if options.split_sentences {
             let texts: Vec<&str> = text.split('\n').collect();
@@ -183,9 +248,10 @@ impl TTSModelHolder {
                 }
                 let (bert_ori, phones, tones, lang_ids) = self.parse_text(t)?;
                 let audio = model::synthesize(
-                    &self.find_model(ident)?.vits2,
+                    vits2,
                     bert_ori.to_owned(),
                     phones,
+                    Array1::from_vec(vec![speaker_id]),
                     tones,
                     lang_ids,
                     style_vector.clone(),
@@ -204,9 +270,10 @@ impl TTSModelHolder {
         } else {
             let (bert_ori, phones, tones, lang_ids) = self.parse_text(text)?;
             model::synthesize(
-                &self.find_model(ident)?.vits2,
+                vits2,
                 bert_ori.to_owned(),
                 phones,
+                Array1::from_vec(vec![speaker_id]),
                 tones,
                 lang_ids,
                 style_vector,
@@ -216,35 +283,6 @@ impl TTSModelHolder {
         };
         tts_util::array_to_vec(audio_array)
     }
-
-    /// Synthesize text to audio
-    ///
-    /// # Note
-    /// This function is for low-level usage, use `easy_synthesize` for high-level usage.
-    #[allow(clippy::too_many_arguments)]
-    pub fn synthesize<I: Into<TTSIdent>>(
-        &self,
-        ident: I,
-        bert_ori: Array2<f32>,
-        phones: Array1<i64>,
-        tones: Array1<i64>,
-        lang_ids: Array1<i64>,
-        style_vector: Array1<f32>,
-        sdp_ratio: f32,
-        length_scale: f32,
-    ) -> Result<Vec<u8>> {
-        let audio_array = model::synthesize(
-            &self.find_model(ident)?.vits2,
-            bert_ori.to_owned(),
-            phones,
-            tones,
-            lang_ids,
-            style_vector,
-            sdp_ratio,
-            length_scale,
-        )?;
-        tts_util::array_to_vec(audio_array)
-    }
 }
 
 /// Synthesize options

sbv2_wasm/pnpm-lock.yaml

@@ -10,14 +10,14 @@ importers:
     dependencies:
       onnxruntime-web:
         specifier: ^1.19.2
-        version: 1.19.2
+        version: 1.20.0
     devDependencies:
       '@biomejs/biome':
         specifier: ^1.9.2
         version: 1.9.4
       '@types/node':
         specifier: ^22.7.4
-        version: 22.7.7
+        version: 22.8.0
       esbuild:
         specifier: ^0.24.0
         version: 0.24.0
@@ -254,8 +254,8 @@ packages:
   '@protobufjs/utf8@1.1.0':
     resolution: {integrity: sha512-Vvn3zZrhQZkkBE8LSuW3em98c0FwgO4nxzv6OdSxPKJIEKY2bGbHn+mhGIPerzI4twdxaP8/0+06HBpwf345Lw==}
 
-  '@types/node@22.7.7':
+  '@types/node@22.8.0':
-    resolution: {integrity: sha512-SRxCrrg9CL/y54aiMCG3edPKdprgMVGDXjA3gB8UmmBW5TcXzRUYAh8EWzTnSJFAd1rgImPELza+A3bJ+qxz8Q==}
+    resolution: {integrity: sha512-84rafSBHC/z1i1E3p0cJwKA+CfYDNSXX9WSZBRopjIzLET8oNt6ht2tei4C7izwDeEiLLfdeSVBv1egOH916hg==}
 
   esbuild@0.24.0:
     resolution: {integrity: sha512-FuLPevChGDshgSicjisSooU0cemp/sGXR841D5LHMB7mTVOmsEHcAxaH3irL53+8YDIeVNQEySh4DaYU/iuPqQ==}
@@ -271,11 +271,11 @@ packages:
   long@5.2.3:
     resolution: {integrity: sha512-lcHwpNoggQTObv5apGNCTdJrO69eHOZMi4BNC+rTLER8iHAqGrUVeLh/irVIM7zTw2bOXA8T6uNPeujwOLg/2Q==}
 
-  onnxruntime-common@1.19.2:
+  onnxruntime-common@1.20.0:
-    resolution: {integrity: sha512-a4R7wYEVFbZBlp0BfhpbFWqe4opCor3KM+5Wm22Az3NGDcQMiU2hfG/0MfnBs+1ZrlSGmlgWeMcXQkDk1UFb8Q==}
+    resolution: {integrity: sha512-9ehS4ul5fBszIcHhfxuDgk45lO+Fqrxmrgwk1Pxb1JRvbQiCB/v9Royv95SRCWHktLMviqNjBsEd/biJhd39cg==}
 
-  onnxruntime-web@1.19.2:
+  onnxruntime-web@1.20.0:
-    resolution: {integrity: sha512-r0ok6KpTUXR4WA+rHvUiZn7JoH02e8iS7XE1p5bXk7q3E0UaRFfYvpMNUHqEPiTBMuIssfBxDCQjUihV8dDFPg==}
+    resolution: {integrity: sha512-IoUf8dqHFJLV4DUSz+Ok+xxyN6cQk57gb20m6PZE5gag3QXuvegYMq9dG8t/QF4JjTKIwvfvnr16ouzCCB9IMA==}
 
   platform@1.3.6:
     resolution: {integrity: sha512-fnWVljUchTro6RiCFvCXBbNhJc2NijN7oIQxbwsyL0buWJPG85v81ehlHI9fXrJsMNgTofEoWIQeClKpgxFLrg==}
@@ -424,7 +424,7 @@ snapshots:
 
   '@protobufjs/utf8@1.1.0': {}
 
-  '@types/node@22.7.7':
+  '@types/node@22.8.0':
     dependencies:
       undici-types: 6.19.8
 
@@ -461,14 +461,14 @@ snapshots:
 
   long@5.2.3: {}
 
-  onnxruntime-common@1.19.2: {}
+  onnxruntime-common@1.20.0: {}
 
-  onnxruntime-web@1.19.2:
+  onnxruntime-web@1.20.0:
     dependencies:
       flatbuffers: 1.12.0
       guid-typescript: 1.0.9
       long: 5.2.3
-      onnxruntime-common: 1.19.2
+      onnxruntime-common: 1.20.0
       platform: 1.3.6
       protobufjs: 7.4.0
 
@@ -486,7 +486,7 @@ snapshots:
       '@protobufjs/path': 1.1.2
       '@protobufjs/pool': 1.1.0
       '@protobufjs/utf8': 1.1.0
-      '@types/node': 22.7.7
+      '@types/node': 22.8.0
       long: 5.2.3
 
   typescript@5.6.3: {}