Changelog

[0.1.4] - 2026年05月07日

Fixed

• Skip test_whisper_probe when flite or openai-whisper is not installed (fixes CI)
• Fixed test_cli_version assertion for v0.1.3

Removed

• Blank demo SVG from README (will be replaced with proper recording later)

[0.1.3] - 2026年05月07日

Fixed

• Made openai-whisper imports lazy so CI/tests pass without the optional [ai] dependency
• Fixed test_cli_version assertion for v0.1.2

[0.1.2] - 2026年05月07日

Fixed

• Fixed author email in pyproject.toml (codinglone@example.com → codinglone@gmail.com)
• Added missing Pydantic validator for speaking_rate_range (length, ordering, non-negative)
• Fixed test assertions to match updated default config values

Added

• GitHub Actions CI workflow running pytest across Python 3.9–3.12
• CI badge in README
• Terminal demo recording (demo/demo.svg) for README
• Issue templates and contributing guidelines

Changed

• NEXT_STEPS.md narrative flipped: deterministic-first, Whisper as optional footnote

[0.1.1] - 2026年05月07日

Changed

• Moved openai-whisper from required to optional dependency (pip install "sonic-gate[ai]")
• Default install is now deterministic-only (fast, no AI overhead)
• Updated PyPI keywords and classifiers for better discoverability
• Added project URLs (Homepage, Repository, Issues, Changelog)

[0.1.0] - 2026年05月07日

Added

• Initial release
• Traditional analyzers: LUFS, silence detection, duration validation, format checking
• Video support with auto audio extraction (MP4, MOV, AVI, MKV, WebM)
• Fix mode: auto-trim silence, normalize LUFS, non-destructive repairs
• Multiple output formats: table, JSON, CSV, markdown
• Optional AI probe via OpenAI Whisper (speech detection, language validation, confidence scoring)
• YAML-based configuration with Pydantic validation
• Typer-based CLI with analyze, demo, and benchmark commands
• Integration test suite (25 tests)

Notes

• AI probe disabled by default after real-world testing showed Whisper is unreliable for low-resource languages (e.g., Kinyarwanda)
• Whisper confidence uses logprob values (negative), not probabilities

Changelog

[0.1.3] - 2026年05月07日

Fixed

• Made openai-whisper imports lazy so CI/tests pass without the optional [ai] dependency
• Fixed test_cli_version assertion for v0.1.2

[0.1.2] - 2026年05月07日

Fixed

• Fixed author email in pyproject.toml (codinglone@example.com → codinglone@gmail.com)
• Added missing Pydantic validator for speaking_rate_range (length, ordering, non-negative)
• Fixed test assertions to match updated default config values

Added

• GitHub Actions CI workflow running pytest across Python 3.9–3.12
• CI badge in README
• Terminal demo recording (demo/demo.svg) for README
• Issue templates and contributing guidelines

Changed

• NEXT_STEPS.md narrative flipped: deterministic-first, Whisper as optional footnote

[0.1.1] - 2026年05月07日

Changed

• Moved openai-whisper from required to optional dependency (pip install "sonic-gate[ai]")
• Default install is now deterministic-only (fast, no AI overhead)
• Updated PyPI keywords and classifiers for better discoverability
• Added project URLs (Homepage, Repository, Issues, Changelog)

[0.1.0] - 2026年05月07日

Added

• Initial release
• Traditional analyzers: LUFS, silence detection, duration validation, format checking
• Video support with auto audio extraction (MP4, MOV, AVI, MKV, WebM)
• Fix mode: auto-trim silence, normalize LUFS, non-destructive repairs
• Multiple output formats: table, JSON, CSV, markdown
• Optional AI probe via OpenAI Whisper (speech detection, language validation, confidence scoring)
• YAML-based configuration with Pydantic validation
• Typer-based CLI with analyze, demo, and benchmark commands
• Integration test suite (25 tests)

Notes

• AI probe disabled by default after real-world testing showed Whisper is unreliable for low-resource languages (e.g., Kinyarwanda)
• Whisper confidence uses logprob values (negative), not probabilities

Changelog

[0.1.2] - 2026年05月07日

Fixed

• Fixed author email in pyproject.toml (codinglone@example.com → codinglone@gmail.com)
• Added missing Pydantic validator for speaking_rate_range (length, ordering, non-negative)
• Fixed test assertions to match updated default config values

Added

• GitHub Actions CI workflow running pytest across Python 3.9–3.12
• CI badge in README
• Terminal demo recording (demo/demo.svg) for README
• Issue templates and contributing guidelines

Changed

• NEXT_STEPS.md narrative flipped: deterministic-first, Whisper as optional footnote

[0.1.1] - 2026年05月07日

Changed

• Moved openai-whisper from required to optional dependency (pip install "sonic-gate[ai]")
• Default install is now deterministic-only (fast, no AI overhead)
• Updated PyPI keywords and classifiers for better discoverability
• Added project URLs (Homepage, Repository, Issues, Changelog)

[0.1.0] - 2026年05月07日

Added

• Initial release
• Traditional analyzers: LUFS, silence detection, duration validation, format checking
• Video support with auto audio extraction (MP4, MOV, AVI, MKV, WebM)
• Fix mode: auto-trim silence, normalize LUFS, non-destructive repairs
• Multiple output formats: table, JSON, CSV, markdown
• Optional AI probe via OpenAI Whisper (speech detection, language validation, confidence scoring)
• YAML-based configuration with Pydantic validation
• Typer-based CLI with analyze, demo, and benchmark commands
• Integration test suite (25 tests)

Notes

• AI probe disabled by default after real-world testing showed Whisper is unreliable for low-resource languages (e.g., Kinyarwanda)
• Whisper confidence uses logprob values (negative), not probabilities

Changelog

[0.1.1] - 2026年05月07日

Changed

• Moved openai-whisper from required to optional dependency (pip install "sonic-gate[ai]")
• Default install is now deterministic-only (fast, no AI overhead)
• Updated PyPI keywords and classifiers for better discoverability
• Added project URLs (Homepage, Repository, Issues, Changelog)

[0.1.0] - 2026年05月07日

Added

• Initial release
• Traditional analyzers: LUFS, silence detection, duration validation, format checking
• Video support with auto audio extraction (MP4, MOV, AVI, MKV, WebM)
• Fix mode: auto-trim silence, normalize LUFS, non-destructive repairs
• Multiple output formats: table, JSON, CSV, markdown
• Optional AI probe via OpenAI Whisper (speech detection, language validation, confidence scoring)
• YAML-based configuration with Pydantic validation
• Typer-based CLI with analyze, demo, and benchmark commands
• Integration test suite (25 tests)

Notes

• AI probe disabled by default after real-world testing showed Whisper is unreliable for low-resource languages (e.g., Kinyarwanda)
• Whisper confidence uses logprob values (negative), not probabilities