v1.7.0

• Add Japanese phonemizer using OpenJTalk (pyopenjtalk-plus) in the new ja extra
  • --data.phoneme_type japanese for training; "phoneme_type": "japanese" in a voice config for synthesis
  • espeak-ng has no kanji coverage (it reads out Unicode character names) and no pitch accent
  • Full-context labels are parsed for pitch accent and mapped to IPA, so Japanese voices stay compatible with the IPA-based (espeak) warmstart
• Add script/setup --ja, and install the ja extra in CI so the Japanese tests run
• libpiper: add piper_create_options and piper_create_with_options(), with piper_create() kept as a wrapper for ABI compatibility

v1.6.1

• Run the g2pW model through piper.g2pw_onnx instead of g2pw.api, dropping torch (~750 MB installed) and requests from the zh extra
  • g2pw.api imports torch only to build padded tensors and iterate batches; the model itself already ran under onnxruntime
  • Also 1.5-2x faster, since it no longer forks DataLoader worker processes on every call
  • g2pW is still required, for its pinyin/bopomofo lookup tables

v1.6.0

• Add Hebrew phonemizer using Nakdimon

v1.5.0

• Add libpiper C++ CLI executable ported from the legacy Piper repository, plus a C++ test suite
• Fix libpiper builds on Windows (MSVC, MSYS2-GCC) and Windows CI
• Bump embedded espeak-ng version
• Add default speaker id for multi-speaker voices
• Add vowel clustering support (--data.vowel_clusters)
• Add in-memory patching for alignments
• Training: add MRD (Multi-Resolution STFT) discriminator, loss/MOS tracking with UTMOS, silence-trim fixes, and dataloader performance improvements
• Pass custom phoneme id map when training

v1.4.2

• Fix pathvalidate dependency

v1.4.1

• Add missing wheels

v1.4.0

• Add Chinese phonemizer based on g2pW
  • Using a quantized version of the original model with quantize_dynamic
• Add --data.phoneme_type pinyin for Chinese phonemization using g2pW
• Add --data.phoneme_type text for using IPA phonemes directly (no espeak-ng)
• Add --model.vocoder_warmstart_ckpt <CHECKPOINT> to restore vocoder params only
• Add --data.dataset_type 'phoneme_ids' to train with pre-generated phoneme ids
  • Use --data.num_symbols <N> to set number of phonemes
  • Use --data.phonemes_path "/path/to/phonemes.json" for phoneme/id map
• Add --output-dir-naming option with timestamp (default) and text

v1.3.0

• Moved development to OHF-Voice org
• Removed C++ code for now to focus on Python development
  • A C API libpiper written in C++ is planned
• Embed espeak-ng directly instead of using separate piper-phonemize library
• Change license to GPLv3
• Use Python stable ABI (3.9+) so only a single wheel per platform is needed
• Change Python API:
  • PiperVoice.synthesize takes a SynthesisConfig and generates AudioChunk objects
  • PiperVoice.synthesize_raw is removed
• Add seperate piper.download_voices utility for downloading voices from HuggingFace
• Allow text as CLI argument: piper ... -- "Text to speak"
• Allow text from one or more files with --input-file <FILE>
• Excluding any file output arguments will play audio directly with ffplay
• Support for raw phonemes in text with [[ <phonemes> ]]
• Adjust output volume with --volume <MULTIPLIER> (default is 1.0)