v0.22.1

What's Changed

Fixes

• Bug fix in MP3 and FLAC compute length on TTSDataset by @Edresson in coqui-ai#3092
• Fix get_git_branch() in generic_utils.py by Nick Potafiy in coqui-ai#3561
• fix(api): handle missing attribute in is_multi_lingual by @eginhard in #4
• Fix TTS().list_models() by @greerviau in #5
• Fix bark model url by @ipeevski in #6
• fix(xtts_manager): name_to_id() should return dict by @eginhard in #8
• fix chinese pinyin phonemes by @JaysonAlbert in #10
• Fix synthesis with fairseq models by @akgupta1337 in #11
• build: update numpy and librosa by @eginhard in #14
• docs(tts.models.vits): clarify use of discriminator/generator by @eginhard in #19
• fix: torch.stft will soon require return_complex=True by @eginhard in #20

Refactors, dependency updates

• Remove pandas dependency by @eginhard in #3
• build: remove unused dependencies by @eginhard in #12
• build: make dependencies for server optional by @eginhard in #15
• Remove duplicate Trainer code by @eginhard in #18
• refactor(dataset): get audio length with torchaudio by @eginhard in #21
• refactor(bin.find_unique_chars): use existing function by @eginhard in #22

Dev improvements

• Fix CI (save best model after 0 steps in tests) by @eginhard in coqui-ai#3423
• CI: remove unused check_skip steps by @akx in #2
• Lint overhaul (pylint to ruff) by @akx in #1
• Pin black for consistent outputs by @eginhard in #7
• test(vocoder): disable wavegrad training test in CI by @eginhard in #9
• ci: combine coverage reports and merge test workflows by @eginhard in #16
• Update readme by @eginhard in #17
• Update repository links, package names, release script by @eginhard in #24

Full Changelog: v0.22.0...v0.22.1