# What changed in sherpa-onnx-core

> 47 changes in sherpa-onnx-core 1 to 2 from its official release notes, upgrade guides and advisories: 2 breaking, 1 security, 18 new APIs and options. Each one cited. Sources checked 2026-10-27. HTML: https://unstale.sh/pypi/sherpa-onnx-core/

To see which of these apply to a repository, run `npx @unstale/cli init` in it.

## sherpa-onnx-core 2

### Breaking changes

- Removes espeak-ng and piper-phonemize from sherpa-onnx. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Raises ValueError when create_stream() cannot encode keywords. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)

### Security

- Fixes a heap overflow for short text at speed greater than 1. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)

### New APIs and options

- Adds an asynchronous creation API to the Dart OfflineRecognizer. (2.0.1) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.1)
- Adds target-speaker voice isolation support and APIs for Python, C/C++, JavaScript, Java, Kotlin, C#, Swift, Go, Dart, Rust, and Pascal. (2.0.1) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.1)
- Supports prompt options per stream for FunASR Nano. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Adds HLG decoding support for offline CTC models across the C, CXX, JNI/Kotlin/Java, and WASM APIs. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Adds the --input-sample-rate option to online-websocket-server. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Supports hotwords with modified_beam_search for streaming NeMo transducer models. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Adds OfflineDiacritization to the Kotlin API. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Adds a Node.js addon for Windows arm64. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Adds GetEmbedding to SpeakerEmbeddingManager. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Ships arm64 libraries in the sherpa_onnx_windows Flutter package. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)

### Behavior and default changes

- Uses librosa-style fbank for streaming NeMo CTC models. (2.0.1) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.1)
- Derives lang2id from the model vocabulary and warns instead of silently falling back to English. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)

### Fixes and other changes

- Fixes decoder positions for generated tokens. (2.0.1) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.1)
- Keeps Kokoro STFT convolutions in float32 when quantizing to int8. (2.0.1) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.1)
- Fixes character loss for byte-level-BPE tokens in whisper and cohere recognizers. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Forwards provider config keys to the CUDA execution provider. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Converts C++ exceptions in recognizer and punctuation constructors to Java exceptions. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Fixes JSON escaping in online and offline recognition results. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Fixes word boundaries, OOV words, and Unicode handling in lexicon TTS. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Feeds the decoder encoder_attention_mask at raw audio length. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Allows the na language tag for Supertonic. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Uses NeMo's additive log-zero guard for NeMo mel features. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Fixes timestamps after filtering special tokens. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Prevents omnilingual ASR from aborting on input shorter than 400 samples. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- C API creation functions catch C++ exceptions and return NULL. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Python methods check embedding size and validate speaker embedding dimensions before releasing the GIL. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Python examples pad 1 second of silence at the endpoint to mirror C++ behavior. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)
- Widens VAD sample positions to int64_t to fix long-stream SIGSEGV. (2.0.0) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v2.0.0)

## sherpa-onnx-core 1

### New APIs and options

- Adds the em_santa voice for Kokoro v1.0. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Adds Go API support for Windows arm64. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Adds Go bindings for speaker diarization confidence computation. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Adds Intel NPU support through OpenVINO. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Adds conversion scripts for running Nabra-82M Arabic TTS with the Kokoro runtime. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Adds per-segment confidence computation for OfflineSpeakerDiarization. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Supports using sherpa-onnx with Python in Termux on Android. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Adds confidence computation support for speaker diarization to JNI. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)

### Tooling

- Supports building within a Termux terminal on Android devices. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)

### Fixes and other changes

- Fixes empty transcripts when the first decoded token is eos. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Fixes hallucinated sentences on silent audio. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Fixes hallucinated text on silent audio. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Fixes hallucinated text on silent audio. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Uploads an int8 model for Nabra-82M Arabic TTS. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Fixes wrapper.py raising NameError because feat_dim is undefined. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)
- Fixes a data race on the shared PRNG in FunASR-Nano and Qwen3-ASR offline recognizers. (1.13.8) — [release notes](https://github.com/k2-fsa/sherpa-onnx/releases/tag/v1.13.8)