kws-models

Refactor the UI of Android TTS engine (#533)

Release v1.9.7

What's Changed

• Replace Android system TTS engine by @csukuangfj in #508
• Build text-to-speech engine APKs by @csukuangfj in #509

Full Changelog: v1.9.5...v1.9.7

Release v1.9.5

What's Changed

• Fix building wheels for Linux. by @csukuangfj in #484
• Fix CI by @csukuangfj in #485
• Print informative error messages for sherpa-onnx-alsa on errors. by @csukuangfj in #486
• Keep multiple threads from calling into espeak-ng at the same time by @csukuangfj in #489
• Fix whisper test script for the latest onnxruntime. by @csukuangfj in #494
• Release Python GIL in C++ class constructor by @csukuangfj in #493
• Support streaming zipformer CTC by @csukuangfj in #496

Full Changelog: v1.9.4...v1.9.5

Release v1.9.4

What's Changed

• Give an informative log for whisper on exceptions. by @csukuangfj in #473
• convert wespeaker models to sherpa-onnx by @csukuangfj in #475
• Fix releasing go packages by @csukuangfj in #476
• Support playing as it is generating for Android by @csukuangfj in #477
• Fix android tts audio buffer size and fix CI. by @csukuangfj in #478
• Add two GLaDOS TTS models by @csukuangfj in #481
• Play generated audio using alsa for TTS by @csukuangfj in #482

Full Changelog: v1.9.1...v1.9.4

Release v1.9.1

What's Changed

• Remove the 30-second constraint from whisper. by @csukuangfj in #471
• Support distil-small.en whisper by @csukuangfj in #472

Full Changelog: v1.9.0...v1.9.1

Speaker recognition models

This release contains speaker recognition models for sherpa-onnx.

Each model has its own license. Please see the corresponding repository for the specific license of a given model.

Release v1.9.0

What's Changed

• Build building for iOS by @csukuangfj in #430
• Judge before UseCachedDecoderOut by @HieDean in #431
• Build MFC examples for Windows x86 (Win32) by @csukuangfj in #434
• Replace Clone() with View() by @HieDean in #432
• Refactor CI scripts about building wheels by @csukuangfj in #436
• support nodejs by @csukuangfj in #438
• Add Swift API for TTS by @csukuangfj in #439
• Text-to-speech for iOS by @csukuangfj in #443
• Lock before push_back the deque for thread safety by @HieDean in #445
• Update to onnxruntime 1.16.3 by @csukuangfj in #446
• Fix reading tokens.txt on Windows by @csukuangfj in #448
• Fix nodejs on Windows by @csukuangfj in #450
• Release GIL to support multithreading in Python websocket servers. by @csukuangfj in #451
• Support piper-phonemize by @csukuangfj in #452
• Use piper-phonemize to convert text to token IDs by @csukuangfj in #453
• Fix CI by @csukuangfj in #456
• Play generated audio as it is generating. by @csukuangfj in #457
• Break text into sentences for tts. by @csukuangfj in #460
• Support playing generated audio as it is generating for MFC. by @csukuangfj in #462
• Fix building for .Net by @csukuangfj in #463
• Use espeak-ng for coqui-ai/TTS VITS English models. by @csukuangfj in #466
• Support Ukrainian VITS models from coqui-ai/TTS by @csukuangfj in #469
• Release v1.9.0 by @csukuangfj in #470

New Contributors

• @HieDean made their first contribution in #431

Full Changelog: v1.8.10...v1.9.0

tts-models

This release contains pre-trained tts models.

Please refer to
https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/index.html
for more models.

Pre-built Android APKs are available at https://k2-fsa.github.io/sherpa/onnx/tts/apk-engine.html

You can try all of the models by visiting the following huggingface space
https://huggingface.co/spaces/k2-fsa/text-to-speech

asr-models

This release contains pre-trained ASR models.

Please refer to
https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html
for more models.

Release v1.8.10

What's Changed

• Fix punctuations in tts for Chinese by @csukuangfj in #417
• Build Android APKs for VITS models from Coqui-ai/TTS by @csukuangfj in #419
• Add a C++ example to show streaming VAD + non-streaming ASR. by @csukuangfj in #420
• Update onnxruntime from v1.16.1 to v1.16.2 by @csukuangfj in #421
• Resize circular buffer on overflow by @csukuangfj in #422
• Add scripts to export ASR models from wenet to ONNX by @csukuangfj in #425
• Support non-streaming WeNet CTC models. by @csukuangfj in #426
• Support streaming conformer CTC models from wenet by @csukuangfj in #427
• Add Python APIs for WeNet CTC models by @csukuangfj in #428

Full Changelog: v1.8.9...v1.8.10

TTS APKs

Please see
https://k2-fsa.github.io/sherpa/onnx/tts/apk.html