mtfyun/sherpa-onnx
1
0 Fork 0
mirror of https://github.com/k2-fsa/sherpa-onnx.git synced 2026-11-10 21:11:03 +08:00
Code Issues Packages Projects Releases Wiki Activity
2,171 Commits 3 Branches 197 Tags
4161dacf2a9a86e3c4dd773b106c557b0a63cc1a
Commit Graph
100 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
Fangjun Kuang 4161dacf2a Fix issues reported by clang-tidy and clang-format (#4130) 2026-10-29 10:30:18 +08:00
Fangjun Kuang 6897d04979 Refactor getting array element length for node-addon (#4129) 2026-10-29 10:23:26 +08:00
Fangjun Kuang 5fc2ced341 Add multi-channel wave IO to the Node.js and HarmonyOS APIs (#4124) 2026-10-28 22:48:47 +08:00
Fangjun Kuang af2d8d81db Release v2.0.1 (#4115) 2026-10-26 22:57:12 +08:00
Fangjun Kuang 96255a1bb2 Add Pascal API for VoxLock (#4114) 2026-10-26 21:50:15 +08:00
Fangjun Kuang 7c5d107813 Add Rust API for VoxLock (#4112) 2026-10-26 21:49:34 +08:00
Fangjun Kuang 436c059fd3 Add Dart API for VoxLock (#4111) 2026-10-26 21:47:08 +08:00
Fangjun Kuang 52d85d3da1 Add Go API for VoxLock (#4110) 2026-10-26 21:36:46 +08:00
Fangjun Kuang 37ce81d5a0 Add Swift API for VoxLock (#4109) 2026-10-26 21:34:12 +08:00
Fangjun Kuang b17952620d Fix CI failures (#4113) 2026-10-26 21:27:40 +08:00
Fangjun Kuang 8b5ff1aa5e Add C# API for VoxLock (#4108) 2026-10-26 19:44:14 +08:00
Fangjun Kuang 687a26126c Add Java and Kotlin API for VoxLock (#4107) 2026-10-26 19:40:49 +08:00
Fangjun Kuang 42e1e0d077 Add JavaScript (WebAssembly) API for VoxLock (#4106) 2026-10-26 19:33:06 +08:00
Fangjun Kuang cb2071f22f Add JavaScript (node-addon) API for VoxLock (#4105) 2026-10-26 18:49:01 +08:00
Fangjun Kuang 1ec9807f1e Add C/CXX API for VoxLock (#4104) 2026-10-26 17:21:30 +08:00
Fangjun Kuang 5f97192eb9 Add Python API for VoxLock (#4102) 2026-10-26 16:31:50 +08:00
Fangjun Kuang 5a26343944 Upload VoxLock models and test data (#4101) 2026-10-26 12:27:43 +08:00
Fangjun Kuang 88204440c2 Fix publishing flutter packages (#4091) 2026-10-23 23:48:11 +08:00
Fangjun Kuang 8abb7c8f87 Release v2.0.0 (#4090) 2026-10-23 20:32:33 +08:00
Fangjun Kuang bad60aa0b5 Fix Ascend CI (#4089) 2026-10-23 20:09:01 +08:00
Fangjun Kuang 12b4f4af2f Add tauri example for Doubao ASR (#4083) 2026-10-23 14:36:20 +08:00
Fangjun Kuang f2e01fe8e3 Widen VAD sample positions to int64_t to fix long-stream SIGSEGV. (#4079) 
Fixes #3635

It is for v2.0.0 and breaks the ABI.

CircularBuffer::head_ and tail_ are absolute sample counters that grow
forever. As int32_t they overflow after 2^31 samples (~37 h at 16 kHz);
tail_ % capacity then goes negative and std::copy() writes to an invalid
iterator, crashing long-running VAD streams.

Widen every absolute stream position to int64_t. Counts and capacities
(Push/Pop/Get's n, Size(), capacity) stay int32_t — they are bounded by
buffer memory, not stream duration.
 2026-10-22 23:39:59 +08:00
Fangjun Kuang 51ec64f7c4 Build wheels for Python 3.15gg (#4080) 2026-10-22 15:17:57 +08:00
Fangjun Kuang a9193276e4 Fix building Flutter APPs for macOS (#4078) 2026-10-22 13:44:57 +08:00
Fangjun Kuang ce949db9dd Fix word boundaries, OOV words and Unicode handling in lexicon TTS (#4075) 2026-10-22 12:16:36 +08:00
Fangjun Kuang a83acee682 Fix typos in the project. (#4073) 2026-10-21 09:57:49 +08:00
Fangjun Kuang bddf3a0118 Refactor building wheels for Linux armv7 (#4071) 2026-10-20 18:44:24 +08:00
Fangjun Kuang db749e7225 Add tests for the windows arm64 npm package (#4069) 2026-10-20 15:42:34 +08:00
Fangjun Kuang 0b3647632f Remove espeak-ng and piper-phonemize from sherpa-onnx (#4055) 
Fixes #3731

espeak-ng and piper-phonemize are no longer built into sherpa-onnx.

Text frontends are now:

- **lexicon** — built in, no extra dependency
- **phonemize** — phonemize outside sherpa-onnx (e.g. [piper-phonemize](https://github.com/csukuangfj/piper-phonemize)) and pass the result in. That library is GPL, so it stays optional: sherpa-onnx is GPL-free after this PR.

## API

- `SherpaOnnxGenerationConfig` gains `phoneme_codepoints` and `tokens` (mutually exclusive alternatives to `text`)
- `data_dir` on the TTS model configs is deprecated and ignored; the ABI slot is kept
- all bindings updated: C/C++, Python, Java, Kotlin, Go, Rust, C#, Swift, Pascal, Dart/Flutter, node-addon, WASM, HarmonyOS
 2026-10-20 11:09:04 +08:00
Fangjun Kuang a452f9f078 Fix Flutter VAD+ASR examples for Qwen3 ASR (#4045) 2026-10-13 20:23:09 +08:00
Fangjun Kuang 9db1af1871 Update README to include deepseek-harness (#4030) 2026-10-09 15:49:02 +08:00
Fangjun Kuang e78dd83ca9 Update README to include China Life (#4029) 2026-10-09 15:29:18 +08:00
Fangjun Kuang f28c1d1840 Update README to include China Unicom (#4028) 2026-10-09 15:06:16 +08:00
Fangjun Kuang 1ca55ff076 Add subtitle generation Python API example for Doubao ASR. (#4027) 
* Add subtitle generation Python API example for Doubao ASR.

generate-subtitles-doubao.py uses silero-vad/ten-vad for segmentation and
the Doubao bigmodel recording-file recognition API for ASR. Audio segments
are sent as base64 encoded wav data, so no public URL is needed. Results
are written to <input>.srt and <input>.txt.

Also for generate-subtitles.py:

 - Add --max-duration-seconds to process only the first N seconds of the
   input file
 - Refuse to run if <input>.srt or <input>.txt already exists
 2026-10-09 13:55:48 +08:00
Fangjun Kuang e20ed70c35 Add subtitle generation Python api example for Qwen3 ASR (#3972) 2026-09-21 16:36:27 +08:00
Fangjun Kuang 275ee49637 Fix dart for windows arm64 (#3967) 2026-09-20 18:03:27 +08:00
Fangjun Kuang e8199a90e1 Add wasm demo for kokoro multi lang tts (#3966) 2026-09-20 14:25:14 +08:00
Fangjun Kuang bc5157a5fd Fix releasing dart packages (#3948) 2026-09-11 14:27:08 +08:00
Fangjun Kuang 11afbd009a Release v1.13.8 (#3945) 2026-09-10 21:52:45 +08:00
Fangjun Kuang 8c81710f76 Test using uv to install sherpa-onnx (#3946) 2026-09-10 21:47:46 +08:00
Fangjun Kuang b0899d95ee Add confidence computation support for speaker diarization to JNI. (#3943) 2026-09-10 17:10:54 +08:00
Fangjun Kuang c7d9eb02b7 Support running Rust examples in termux on Android (#3942) 2026-09-10 15:49:47 +08:00
Fangjun Kuang 6f5327bad8 Add robust download retry logic to all test scripts (#3941) 2026-09-09 17:56:46 +08:00
Fangjun Kuang bf690ea1c7 Update onnxruntime to v1.28.2 (#3935) 2026-09-09 15:21:14 +08:00
Fangjun Kuang a4d0341b42 Fix CI tests (#3940) 2026-09-09 14:57:29 +08:00
Fangjun Kuang fe56f3dec6 Add Go API for Windows arm64 (#3934) 2026-09-08 18:44:27 +08:00
Fangjun Kuang 880f1a8d13 Upload the nabra 82M Arabic TTS int8 model- #3933 (#3933) 2026-09-08 18:43:23 +08:00
Fangjun Kuang 210f340bcf Fix Go API tests in CI for Windows (#3932) 2026-09-08 12:02:57 +08:00
Fangjun Kuang df3a263607 Support sherpa-onnx with Python in Termux on Android (#3922) 2026-09-06 07:02:20 +08:00
Fangjun Kuang cb2beea82b Support building within a termux terminal on Android devices (#3916) 2026-09-04 18:04:44 +08:00
Fangjun Kuang a24dad69b4 Fix releasing (#3911) 2026-09-02 19:23:03 +08:00
Fangjun Kuang 917bed95c8 Release v1.13.7 (#3909) 2026-09-01 12:48:05 +08:00
Fangjun Kuang 0fbdb4c5bd Remove caret prefix from sherpa_onnx flutter dependencies (#3905) 
* Remove caret prefix from sherpa_onnx flutter dependencies

Changed all sherpa_onnx flutter dependencies in pubspec.yaml files
from caret version (^1.13.6) to exact version (1.13.6) matching.
 2026-08-30 11:12:30 +08:00
Fangjun Kuang 055046b25a Add iOS support for Rust and add hello world tauri example (#3901) 2026-08-28 19:47:24 +08:00
Fangjun Kuang 34eba5a272 Fix building APKs in CI (#3896) 2026-08-25 09:44:54 +08:00
Fangjun Kuang d33f87c8ad Support Windows arm64 for Rust (#3889) 2026-08-24 19:14:47 +08:00
Fangjun Kuang 5875a75996 Fix Flutter packages for iOS (#3887) 2026-08-24 14:35:41 +08:00
Fangjun Kuang 1cb484af5e Release v1.13.6 (#3879) 2026-08-18 13:59:39 +08:00
Fangjun Kuang 0cefb05e7d Add VAD+ASR Flutter examples (#3877) 2026-08-18 11:28:11 +08:00
Fangjun Kuang 8c0c1b02ca Unify sherpa-onnx init API for Flutter and Dart CLI (#3875) 2026-08-13 16:53:14 +08:00
Fangjun Kuang e1edbfee66 Fix releasing dart packages (#3874) 2026-08-13 09:43:57 +08:00
Fangjun Kuang c3625d24fb Fix using SPM in iOS examples (#3867) 2026-08-11 20:06:04 +08:00
Fangjun Kuang 99dcbfe760 Fix building android and java examples (#3866) 2026-08-11 17:54:38 +08:00
Fangjun Kuang 3dc7c569f3 Release v1.13.5 (#3863) 2026-08-11 14:43:30 +08:00
Fangjun Kuang 952b37032f Fix float32 catastrophic cancellation in variance computation (#3862) 
Compute variance from centered values (x - mean)^2 instead of
E[x^2] - E[x]^2 in ComputeMeanAndInvStd, Omnilingual ASR CTC
normalization, and SpeakerEmbeddingExtractorNeMo NormalizePerFeature.

The subtraction E[x^2] - E[x]^2 cancels catastrophically in float32
when a feature stays near a constant value, yielding var == 0 and
huge normalized outliers.

Applies the same fix as PR #3857 (NemoNormalizePerFeature) to all
remaining instances of this pattern.
 2026-08-11 11:32:59 +08:00
Fangjun Kuang 02b3e424ba Update onnxruntime to v1.27.1 (#3861) 2026-08-11 11:14:09 +08:00
Fangjun Kuang a04a27dcf8 Fix Kotlin Syntax errors (#3860) 2026-08-11 10:33:38 +08:00
Fangjun Kuang 17ab05b760 Add Flutter examples for VAD (#3856) 2026-08-10 21:00:44 +08:00
Fangjun Kuang 6897144f08 Add CI for Flutter punctuation examples (#3847) 2026-08-07 17:48:34 +08:00
Fangjun Kuang ceec1868fa Add punctuation examples for Flutter (#3846) 2026-08-07 14:33:29 +08:00
Fangjun Kuang fa76497c37 Build libs and Python wheels for CUDA 13 (#3844) 2026-08-07 11:53:28 +08:00
Fangjun Kuang f737aebe7c Add Flutter/Dart Web support via WebAssembly (#3842) 2026-08-06 17:10:10 +08:00
Fangjun Kuang dcf56735d4 Disable pthread support in WebAssembly. (#3836) 
It removes WebAssembly pthread usage across the WASM demos/builds (simplifying runtime requirements) and updates the wasm SIMD onnxruntime package to 1.27.1. It also adjusts the JS worker/loading strategy and Node.js packaging to match the single-threaded WASM approach.
 2026-08-04 10:06:58 +08:00
Fangjun Kuang ba07953c2f Fix building Python wheels on macOS (#3837) 
We have removed the `soversion` name from the lib name since `onnxruntime` 1.27.1, so there is no `libonnxruntime.x.y.z.dylib` anymore.
 2026-08-03 19:39:21 +08:00
Fangjun Kuang c0c00c9b5a Support using SPM for Flutter on macOS and iOS (#3835) 2026-08-03 18:03:26 +08:00
Fangjun Kuang c29b1838c8 Support static linking onnxruntime in xcframework with shared lib. (#3833) 2026-08-03 13:00:17 +08:00
Fangjun Kuang 2e2659e937 Refactor building xcframework for macOS and iOS (#3831) 2026-08-03 10:32:47 +08:00
Fangjun Kuang 116a44e72c Set onnxruntime version to 1.27.1 in SPM (#3830) 2026-07-30 17:04:11 +08:00
Fangjun Kuang 5e4232db78 Add missing files for swift package manager (#3828) 2026-07-30 11:46:54 +08:00
Fangjun Kuang 9ad8c8891e Fix iOS examples using swift package manager (#3823) 2026-07-29 17:56:49 +08:00
Fangjun Kuang b5f5866dae Use SPM in iOS examples (#3822) 2026-07-29 15:31:20 +08:00
Fangjun Kuang 14190197d9 Support swift package manager (SPM) (#3816) 2026-07-29 10:04:08 +08:00
Fangjun Kuang 75e1fc31e7 Build Python wheels for Windows arm64 (#3812) 2026-07-28 17:05:19 +08:00
Fangjun Kuang 282e20c846 Add Android demo for Inflect TTS (#3811) 2026-07-28 17:04:22 +08:00
Fangjun Kuang ccdaef8bba Add API to get onnxruntime version. (#3810) 2026-07-28 14:02:43 +08:00
Fangjun Kuang 0a03d8546f Export Inflect TTS models to sherpa-onnx (#3808) 2026-07-27 20:36:04 +08:00
Fangjun Kuang 546df6f963 Test Java API on Windows arm64 (#3801) 2026-07-24 19:11:20 +08:00
Fangjun Kuang 881ce3078f Simplify build.gradle.kts for java api examples. (#3788) 2026-07-24 17:01:43 +08:00
Fangjun Kuang 24a84a910a Refactor building and testing Java API (#3786) 2026-07-23 14:12:31 +08:00
Fangjun Kuang 2cbcb26887 Fix greedy search decoding for Nemo streaming transducer (#3785) 2026-07-23 11:16:00 +08:00
Fangjun Kuang 9d15a282aa Add Java API examples with gradle (#3784) 2026-07-22 19:36:59 +08:00
Fangjun Kuang a3b58a159e Add maven examples for Java API (#3783) 2026-07-22 18:47:01 +08:00
Fangjun Kuang fd63389bb4 Fix uploading Piper TTS models (#3774) 2026-07-20 10:02:25 +08:00
Fangjun Kuang 8e7ca3cdfa Upload vits-piper-pt_BR-cadu-medium to huggingface (#3763) 2026-07-15 16:55:00 +08:00
Fangjun Kuang 7fc2ea7725 Add C++ runtime for Moonshine ASR with QNN for Qualcomm NPU (#3758) 2026-07-14 14:36:02 +08:00
Fangjun Kuang d5c2ab3c36 Export Moonshine ASR models to QNN (#3756) 2026-07-13 16:12:29 +08:00
Fangjun Kuang 97293f0d9b Fix build warnings (#3742) 2026-07-09 19:25:44 +08:00
Fangjun Kuang 9b3374251b Export nemotron-3.5-asr-streaming-0.6b to QNN (#3741) 2026-07-09 18:01:21 +08:00
Fangjun Kuang f8ffa43055 Build docker image to setup the QNN SDK environment (#3736) 2026-07-08 22:24:49 +08:00
Fangjun Kuang 6a204636c4 Refactor ONNX export script for nemotron-3.5-asr-streaming-0.6b (#3734) 2026-07-08 18:07:41 +08:00