echogarden 1.6.2 → 1.8.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/data/schemas/options.json +50 -12
- package/dist/api/Alignment.js +2 -2
- package/dist/api/Alignment.js.map +1 -1
- package/dist/api/Denoising.js +5 -4
- package/dist/api/Denoising.js.map +1 -1
- package/dist/api/Recognition.js +2 -2
- package/dist/api/Recognition.js.map +1 -1
- package/dist/api/SourceSeparation.d.ts +2 -5
- package/dist/api/SourceSeparation.js +14 -13
- package/dist/api/SourceSeparation.js.map +1 -1
- package/dist/api/SpeechLanguageDetection.js +2 -2
- package/dist/api/SpeechLanguageDetection.js.map +1 -1
- package/dist/api/SpeechTranslation.js +2 -2
- package/dist/api/SpeechTranslation.js.map +1 -1
- package/dist/api/Synthesis.js +8 -6
- package/dist/api/Synthesis.js.map +1 -1
- package/dist/api/TextTranslation.d.ts +3 -1
- package/dist/api/TextTranslation.js +12 -6
- package/dist/api/TextTranslation.js.map +1 -1
- package/dist/api/TranslationAlignment.js +2 -2
- package/dist/api/TranslationAlignment.js.map +1 -1
- package/dist/audio/AudioUtilities.d.ts +5 -3
- package/dist/audio/AudioUtilities.js +39 -25
- package/dist/audio/AudioUtilities.js.map +1 -1
- package/dist/build-tools/MakeTarballsForInstalledPackages.d.ts +1 -0
- package/dist/build-tools/MakeTarballsForInstalledPackages.js +20 -0
- package/dist/build-tools/MakeTarballsForInstalledPackages.js.map +1 -0
- package/dist/cli/CLI.js +4 -3
- package/dist/cli/CLI.js.map +1 -1
- package/dist/dsp/FFT.d.ts +3 -3
- package/dist/dsp/FFT.js +55 -34
- package/dist/dsp/FFT.js.map +1 -1
- package/dist/nlp/IPA.d.ts +3 -3
- package/dist/nlp/Segmentation.js +2 -1
- package/dist/nlp/Segmentation.js.map +1 -1
- package/dist/nlp/TextNormalizer.d.ts +2 -0
- package/dist/nlp/TextNormalizer.js +51 -0
- package/dist/nlp/TextNormalizer.js.map +1 -1
- package/dist/recognition/WhisperSTT.d.ts +5 -0
- package/dist/recognition/WhisperSTT.js +19 -11
- package/dist/recognition/WhisperSTT.js.map +1 -1
- package/dist/source-separation/MDXNetSourceSeparation.d.ts +26 -3
- package/dist/source-separation/MDXNetSourceSeparation.js +159 -79
- package/dist/source-separation/MDXNetSourceSeparation.js.map +1 -1
- package/dist/subtitles/Subtitles.js +2 -1
- package/dist/subtitles/Subtitles.js.map +1 -1
- package/dist/synthesis/EspeakTTS.js +5 -5
- package/dist/synthesis/EspeakTTS.js.map +1 -1
- package/dist/tests/Test.js +0 -1
- package/dist/tests/Test.js.map +1 -1
- package/dist/text-translation/GoogleTranslateTextTranslation.d.ts +9 -3
- package/dist/text-translation/GoogleTranslateTextTranslation.js +59 -29
- package/dist/text-translation/GoogleTranslateTextTranslation.js.map +1 -1
- package/dist/utilities/OnnxUtilities.d.ts +1 -0
- package/dist/utilities/OnnxUtilities.js +5 -5
- package/dist/utilities/OnnxUtilities.js.map +1 -1
- package/dist/utilities/PackageManager.js +6 -6
- package/dist/utilities/PackageManager.js.map +1 -1
- package/dist/utilities/StringUtilities.d.ts +28 -0
- package/dist/utilities/StringUtilities.js +150 -0
- package/dist/utilities/StringUtilities.js.map +1 -0
- package/dist/utilities/TarballMaker.d.ts +3 -4
- package/dist/utilities/TarballMaker.js +6 -19
- package/dist/utilities/TarballMaker.js.map +1 -1
- package/dist/utilities/Timeline.js +2 -1
- package/dist/utilities/Timeline.js.map +1 -1
- package/dist/utilities/Utilities.d.ts +1 -33
- package/dist/utilities/Utilities.js +3 -176
- package/dist/utilities/Utilities.js.map +1 -1
- package/dist/utilities/WasmMemoryManager.d.ts +9 -1
- package/dist/utilities/WasmMemoryManager.js +24 -3
- package/dist/utilities/WasmMemoryManager.js.map +1 -1
- package/dist/utilities/WebReader.js +1 -1
- package/dist/utilities/WebReader.js.map +1 -1
- package/dist/utilities/WikipediaReader.js +1 -1
- package/dist/utilities/WikipediaReader.js.map +1 -1
- package/dist/utilities/WindowedList.d.ts +10 -0
- package/dist/utilities/WindowedList.js +36 -0
- package/dist/utilities/WindowedList.js.map +1 -0
- package/docs/Engines.md +1 -1
- package/docs/Options.md +16 -10
- package/package.json +11 -11
- package/src/alignment/SpeechAlignment.ts +1 -1
- package/src/api/Alignment.ts +2 -2
- package/src/api/Denoising.ts +5 -4
- package/src/api/Recognition.ts +3 -3
- package/src/api/SourceSeparation.ts +20 -19
- package/src/api/SpeechLanguageDetection.ts +2 -2
- package/src/api/SpeechTranslation.ts +2 -2
- package/src/api/Synthesis.ts +8 -6
- package/src/api/TextTranslation.ts +17 -9
- package/src/api/TranslationAlignment.ts +2 -2
- package/src/audio/AudioUtilities.ts +53 -33
- package/src/build-tools/MakeTarballsForInstalledPackages.ts +25 -0
- package/src/cli/CLI.ts +4 -3
- package/src/dsp/FFT.ts +65 -42
- package/src/nlp/Segmentation.ts +2 -1
- package/src/nlp/TextNormalizer.ts +60 -0
- package/src/recognition/WhisperSTT.ts +25 -12
- package/src/source-separation/MDXNetSourceSeparation.ts +218 -88
- package/src/subtitles/Subtitles.ts +2 -1
- package/src/synthesis/EspeakTTS.ts +5 -5
- package/src/tests/Test.ts +0 -4
- package/src/text-translation/GoogleTranslateTextTranslation.ts +91 -30
- package/src/typings/Fillers.d.ts +2 -0
- package/src/utilities/OnnxUtilities.ts +7 -8
- package/src/utilities/PackageManager.ts +6 -7
- package/src/utilities/StringUtilities.ts +203 -0
- package/src/utilities/TarballMaker.ts +6 -23
- package/src/utilities/Timeline.ts +2 -1
- package/src/utilities/Utilities.ts +4 -237
- package/src/utilities/WasmMemoryManager.ts +38 -4
- package/src/utilities/WebReader.ts +1 -1
- package/src/utilities/WikipediaReader.ts +1 -1
- package/src/utilities/WindowedList.ts +45 -0
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"Segmentation.js","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,gBAAgB,MAAM,mBAAmB,CAAA;AACrD,OAAO,EAAE,6BAA6B,EAAE,MAAM,0BAA0B,CAAA;AAExE,OAAO,EAAE,QAAQ,EAAE,
|
|
1
|
+
{"version":3,"file":"Segmentation.js","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,gBAAgB,MAAM,mBAAmB,CAAA;AACrD,OAAO,EAAE,6BAA6B,EAAE,MAAM,0BAA0B,CAAA;AAExE,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACjE,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAC7D,OAAO,EAAE,iCAAiC,EAAE,MAAM,2BAA2B,CAAA;AAE7E,OAAO,EAAE,aAAa,EAAE,YAAY,EAAE,MAAM,iCAAiC,CAAA;AAE7E,MAAM,GAAG,GAAG,WAAW,CAAA;AAEvB,MAAM,CAAC,MAAM,oBAAoB,GAAG,yBAAyB,CAAA;AAC7D,MAAM,CAAC,MAAM,kBAAkB,GAAG,oBAAoB,CAAA;AAEtD,MAAM,CAAC,MAAM,gBAAgB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AACzD,MAAM,CAAC,MAAM,WAAW,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAEzI,MAAM,UAAU,kBAAkB,CAAC,GAAW;IAC7C,OAAO,MAAM,CAAC,GAAG,CAAC,IAAI,WAAW,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAA;AAChD,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,GAAW;IACvC,OAAO,WAAW,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AACxC,CAAC;AAED,MAAM,UAAU,MAAM,CAAC,GAAW;IACjC,OAAO,oBAAoB,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC7C,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,GAAW;IACxC,OAAO,kBAAkB,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC3C,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,GAAW;IACvC,OAAO,GAAG,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,CAAA;AAC/B,CAAC;AAED,MAAM,OAAO,QAAQ;IACpB,OAAO,GAAa,EAAE,CAAA;IAEb,mBAAmB,GAAG,IAAI,CAAA;IAEnC,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEzE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,OAAO,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,MAAM,EAAE,EAAE,CAAC,MAAM,GAAG,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;CAC/F;AAED,MAAM,OAAO,MAAM;IAClB,KAAK,GAAW,EAAE,CAAA;IAElB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEnE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,KAAK,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,MAAM,GAAG,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAEzF,IAAI,QAAQ;QACX,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;YAC5B,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IACzC,CAAC;IAED,IAAI,mBAAmB,KAAK,OAAO,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,mBAAmB,CAAC,CAAC,CAAC,KAAK,CAAA,CAAC,CAAC;CACtG;AAED,MAAM,OAAO,IAAI;IACP,IAAI,CAAQ;IACrB,mBAAmB,CAAS;IAE5B,YAAY,IAAY,EAAE,mBAA4B;QACrD,IAAI,CAAC,IAAI,GAAG,IAAI,CAAA;QAChB,IAAI,CAAC,mBAAmB,GAAG,mBAAmB,CAAA;IAC/C,CAAC;IAED,IAAI,uBAAuB,KAAK,OAAO,CAAC,oBAAoB,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,YAAY,CAAA,CAAC,CAAC;IAEpG,IAAI,YAAY,KAAK,OAAO,WAAW,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA,CAAC,CAAC;IAE7D,IAAI,iBAAiB,KAAK,OAAO,IAAI,CAAC,uBAAuB,IAAI,aAAa,CAAC,IAAI,CAAC,IAAI,EAAE,gBAAgB,CAAC,CAAA,CAAC,CAAC;IAE7G,IAAI,MAAM,KAAK,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,CAAA,CAAC,CAAC;CACxC;AAID,MAAM,OAAO,QAAQ;IACpB,QAAQ,GAAc,EAAE,CAAA;IAExB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAE1E,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,OAAO,EAAE,EAAE,CAAC,MAAM,GAAG,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAElG,IAAI,OAAO,KAAK,OAAO,IAAI,CAAC,MAAM,IAAI,CAAC,CAAA,CAAC,CAAC;IAEzC,IAAI,UAAU,KAAK,OAAO,CAAC,IAAI,CAAC,OAAO,CAAA,CAAC,CAAC;IAEzC,IAAI,WAAW;QACd,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YAClB,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAC/C,CAAC;CACD;AAED,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,IAAY,EAAE,iBAAyB,EAAE,QAAgB,EAAE,iBAAiB,GAAG,IAAI,EAAE,eAAe,GAAG,IAAI;IACjJ,MAAM,UAAU,GAAG,MAAM,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IAE9C,MAAM,SAAS,GAAe,EAAE,CAAA;IAChC,IAAI,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;IAEpC,MAAM,oCAAoC,GAAG,GAAG,EAAE,CAAC,iBAAiB,GAAG,eAAe,CAAC,MAAM,CAAA;IAC7F,MAAM,yBAAyB,GAAG,GAAG,EAAE;QACtC,IAAI,eAAe,CAAC,UAAU,EAAE,CAAC;YAChC,SAAS,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;YAC/B,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;QACjC,CAAC;IACF,CAAC,CAAA;IAED,MAAM,mBAAmB,GAAG,CAAC,OAAgB,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,IAAI,oCAAoC,EAAE,CAAA;IAE1G,KAAK,MAAM,QAAQ,IAAI,UAAU,EAAE,CAAC;QACnC,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;YACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YACvC,SAAQ;QACT,CAAC;QAED,IAAI,iBAAiB,EAAE,CAAC;YACvB,yBAAyB,EAAE,CAAA;YAE3B,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;gBACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;gBACvC,SAAQ;YACT,CAAC;QACF,CAAC;QAED,KAAK,MAAM,MAAM,IAAI,QAAQ,CAAC,OAAO,EAAE,CAAC;YACvC,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;gBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;gBACrC,SAAQ;YACT,CAAC;YAGD,IAAI,eAAe,EAAE,CAAC;gBACrB,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;oBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;oBACrC,SAAQ;gBACT,CAAC;YACF,CAAC;YAED,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,KAAK,EAAE,CAAC;gBACjC,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,MAAM,IAAI,KAAK,CAAC,gCAAgC,IAAI,CAAC,MAAM,mDAAmD,iBAAiB,EAAE,CAAC,CAAA;YACnI,CAAC;QACF,CAAC;IACF,CAAC;IAED,yBAAyB,EAAE,CAAA;IAE3B,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,KAAK,CAAC,IAAY,EAAE,QAAgB;IACzD,MAAM,aAAa,GAAG,gBAAgB,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IACtD,MAAM,SAAS,GAAe,EAAE,CAAA;IAEhC,KAAK,MAAM,YAAY,IAAI,aAAa,EAAE,CAAC;QAC1C,MAAM,QAAQ,GAAG,IAAI,QAAQ,EAAE,CAAA;QAE/B,IAAI,aAAa,GAAG,IAAI,MAAM,EAAE,CAAA;QAChC,MAAM,SAAS,GAAG,MAAM,YAAY,CAAC,YAAY,EAAE,QAAQ,CAAC,CAAA;QAE5D,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,SAAS,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,CAAC;YACnE,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,SAAS,CAAC,SAAS,CAAC,EAAE,SAAS,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;YAE9E,IAAI,IAAI,CAAC,iBAAiB,EAAE,CAAC;gBAC5B,MAAM,cAAc,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,EAAE,gBAAgB,CAAC,CAAA;gBAChE,aAAa,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,cAAc,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,mBAAmB,CAAC,CAAC,CAAA;gBACxG,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,aAAa,CAAC,CAAA;gBAEpC,aAAa,GAAG,IAAI,MAAM,EAAE,CAAA;gBAC5B,aAAa,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,cAAc,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAA;YACnF,CAAC;iBAAM,CAAC;gBACP,aAAa,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC/B,CAAC;QACF,CAAC;QAED,IAAI,aAAa,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACpC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,aAAa,CAAC,CAAA;QACrC,CAAC;QAED,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;IACzB,CAAC;IAED,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,IAAY,EAAE,QAAgB;IAC9D,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,OAAO,gBAAgB,CAAC,aAAa,CAAC,IAAI,EAAE,gBAAgB,CAAC,YAAY,CAAC,aAAa,CAAC,CAAC,CAAA;AAC1F,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,IAAY,EAAE,QAAgB;IAChE,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,IAAI,aAAa,IAAI,IAAI,IAAI,aAAa,IAAI,KAAK,EAAE,CAAC;QACrD,OAAO,6BAA6B,CAAC,IAAI,EAAE,SAAS,EAAE,IAAI,CAAC,CAAA;IAC5D,CAAC;SAAM,IAAI,aAAa,IAAI,IAAI,EAAE,CAAC;QAClC,OAAO,iCAAiC,CAAC,IAAI,CAAC,CAAA;IAC/C,CAAC;SAAM,CAAC;QACP,OAAO,gBAAgB,CAAC,SAAS,CAAC,IAAI,EAAE,gBAAgB,CAAC,YAAY,CAAC,aAAa,CAAC,CAAC,CAAA;IACtF,CAAC;AACF,CAAC;AAED,MAAM,UAAU,iBAAiB,CAAC,IAAY,EAAE,eAAmC,EAAE,0BAAgD;IACpI,IAAI,UAAU,GAAa,EAAE,CAAA;IAE7B,IAAI,eAAe,KAAK,QAAQ,EAAE,CAAC;QAClC,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,CAAA;IACrC,CAAC;SAAM,IAAI,eAAe,KAAK,QAAQ,EAAE,CAAC;QACzC,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,kBAAkB,CAAC,CAAA;IAC5C,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,kCAAkC,eAAe,GAAG,CAAC,CAAA;IACtE,CAAC;IAED,UAAU,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,yBAAyB,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,0BAA0B,CAAC,CAAC,CAAA;IACjG,UAAU,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAEjD,OAAO,UAAU,CAAA;AAClB,CAAC;AAED,MAAM,UAAU,yBAAyB,CAAC,IAAY,EAAE,0BAAgD;IACvG,IAAI,0BAA0B,KAAK,kBAAkB,EAAE,CAAC;QACvD,OAAO,IAAI,CAAC,UAAU,CAAC,WAAW,EAAE,GAAG,CAAC,CAAA;IACzC,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC,CAAA;IACpC,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAA;IACZ,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,0CAA0C,0BAA0B,GAAG,CAAC,CAAA;IACzF,CAAC;AACF,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,IAAY;IACxC,OAAO,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAA;AAC5B,CAAC"}
|
|
@@ -4,3 +4,5 @@ export declare function getNormalizedFragmentsForSpeech(words: string[], languag
|
|
|
4
4
|
};
|
|
5
5
|
export declare function normalizeFourDigitYearString(yearString: string): string;
|
|
6
6
|
export declare function normalizeFourDigitDecadeString(decadeString: string): string;
|
|
7
|
+
export declare function simplifyPunctuationCharacters(text: string): string;
|
|
8
|
+
export declare const punctuationSubstitutionLookup: Record<string, string>;
|
|
@@ -1,4 +1,5 @@
|
|
|
1
1
|
import { getShortLanguageCode } from '../utilities/Locale.js';
|
|
2
|
+
import { substituteCharactersUsingLookup } from '../utilities/StringUtilities.js';
|
|
2
3
|
export function getNormalizedFragmentsForSpeech(words, language) {
|
|
3
4
|
language = getShortLanguageCode(language);
|
|
4
5
|
if (language != 'en') {
|
|
@@ -123,4 +124,54 @@ export function normalizeFourDigitDecadeString(decadeString) {
|
|
|
123
124
|
}
|
|
124
125
|
return normalizedString;
|
|
125
126
|
}
|
|
127
|
+
export function simplifyPunctuationCharacters(text) {
|
|
128
|
+
return substituteCharactersUsingLookup(text, punctuationSubstitutionLookup);
|
|
129
|
+
}
|
|
130
|
+
export const punctuationSubstitutionLookup = {
|
|
131
|
+
'“': `"`,
|
|
132
|
+
'”': `"`,
|
|
133
|
+
'‟': `"`,
|
|
134
|
+
'ˮ': `"`,
|
|
135
|
+
'„': `"`,
|
|
136
|
+
'‹': `"`,
|
|
137
|
+
'›': `"`,
|
|
138
|
+
'❮': `"`,
|
|
139
|
+
'❯': '"',
|
|
140
|
+
'«': `"`,
|
|
141
|
+
'»': `"`,
|
|
142
|
+
'״': `"`,
|
|
143
|
+
'❝': `"`,
|
|
144
|
+
'❞': `"`,
|
|
145
|
+
'🙶': `"`,
|
|
146
|
+
'🙷': `"`,
|
|
147
|
+
'⹂': `"`,
|
|
148
|
+
'〝': `"`,
|
|
149
|
+
'〞': `"`,
|
|
150
|
+
'〟': `"`,
|
|
151
|
+
'"': `"`,
|
|
152
|
+
'❠': `"`,
|
|
153
|
+
'🙸': `"`,
|
|
154
|
+
'ߵ': `'`,
|
|
155
|
+
'ߴ': `'`,
|
|
156
|
+
'’': `'`,
|
|
157
|
+
'‘': `'`,
|
|
158
|
+
'ʹ': `'`,
|
|
159
|
+
'ʼ': `'`,
|
|
160
|
+
''': `'`,
|
|
161
|
+
'ʻ': `'`,
|
|
162
|
+
'՚': `'`,
|
|
163
|
+
'՛': `'`,
|
|
164
|
+
'❛': `'`,
|
|
165
|
+
'❜': `'`,
|
|
166
|
+
'❟': `'`,
|
|
167
|
+
',': `,`,
|
|
168
|
+
'、': `,`,
|
|
169
|
+
':': `:`,
|
|
170
|
+
';': `;`,
|
|
171
|
+
'。': `.`,
|
|
172
|
+
'?': `?`,
|
|
173
|
+
'؟': `?`,
|
|
174
|
+
'!': `!`,
|
|
175
|
+
'¡': `!`,
|
|
176
|
+
};
|
|
126
177
|
//# sourceMappingURL=TextNormalizer.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"TextNormalizer.js","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;
|
|
1
|
+
{"version":3,"file":"TextNormalizer.js","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAC7D,OAAO,EAAE,+BAA+B,EAAE,MAAM,iCAAiC,CAAA;AAEjF,MAAM,UAAU,+BAA+B,CAAC,KAAe,EAAE,QAAgB;IAChF,QAAQ,GAAG,oBAAoB,CAAC,QAAQ,CAAC,CAAA;IAEzC,IAAI,QAAQ,IAAI,IAAI,EAAE,CAAC;QACtB,OAAO,EAAE,mBAAmB,EAAE,CAAC,GAAG,KAAK,CAAC,EAAE,kBAAkB,EAAE,CAAC,GAAG,KAAK,CAAC,EAAE,CAAA;IAC3E,CAAC;IAED,MAAM,aAAa,GAAG,mBAAmB,CAAA;IAEzC,MAAM,oBAAoB,GAAG,wBAAwB,CAAA;IACrD,MAAM,sBAAsB,GAAG,qBAAqB,CAAA;IAEpD,MAAM,yBAAyB,GAAG,kDAAkD,CAAA;IAEpF,MAAM,mBAAmB,GAAG;QAC3B,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,IAAI,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO;QACpG,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM;QACjF,SAAS,EAAE,UAAU,EAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,WAAW,EAAE,SAAS,EAAE,UAAU,EAAE,UAAU;QACxH,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK;KAClF,CAAA;IAED,MAAM,qBAAqB,GAAG;QAC7B,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG;KACxC,CAAA;IAED,MAAM,yBAAyB,GAAG;QACjC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG;KAClB,CAAA;IAED,MAAM,gCAAgC,GAAG;QACxC,SAAS,EAAE,OAAO,EAAE,QAAQ,EAAE,KAAK;KACnC,CAAA;IAED,MAAM,wBAAwB,GAAG;QAChC,SAAS,EAAE,SAAS,EAAE,UAAU;KAChC,CAAA;IAED,MAAM,mBAAmB,GAAa,EAAE,CAAA;IACxC,MAAM,kBAAkB,GAAa,EAAE,CAAA;IAEvC,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,KAAK,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,CAAC;QAC/D,MAAM,IAAI,GAAG,KAAK,CAAC,SAAS,CAAC,CAAA;QAC7B,MAAM,aAAa,GAAG,IAAI,CAAC,WAAW,EAAE,CAAA;QAExC,MAAM,SAAS,GAAG,KAAK,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAA;QAC5C,MAAM,QAAQ,GAAG,SAAS,CAAC,CAAC,CAAC,CAAA;QAE7B,KAAK,uDAAuD;QAC3D,mBAAmB,CAAC,QAAQ,CAAC,aAAa,CAAC;YAC3C,oBAAoB,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC;YAEtC,MAAM,gBAAgB,GAAG,4BAA4B,CAAC,QAAQ,CAAC,CAAA;YAE/D,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAE7B,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAC1C,kBAAkB,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YAEjC,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,KAAK,2DAA2D;QACtE,qBAAqB,CAAC,QAAQ,CAAC,aAAa,CAAC;YAC7C,sBAAsB,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC;YAExC,MAAM,gBAAgB,GAAG,8BAA8B,CAAC,QAAQ,CAAC,CAAA;YAEjE,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAE7B,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAC1C,kBAAkB,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YAEjC,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,KAAK,mDAAmD;QAC9D,yBAAyB,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,SAAS,EAAE,SAAS,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC;YAEjF,mBAAmB,CAAC,IAAI,CAAC,4BAA4B,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,CAAA;YACxE,kBAAkB,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,CAAA;YAEzC,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAA;YAE7C,mBAAmB,CAAC,IAAI,CAAC,4BAA4B,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;YAC5E,kBAAkB,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAA;YAE7C,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,KAAK,8DAA8D;QACzE,yBAAyB,CAAC,QAAQ,CAAC,aAAa,CAAC;YACjD,aAAa,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC;YAE/B,IAAI,YAAY,GAAG,gCAAgC,CAAC,yBAAyB,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC,CAAA;YAErG,IAAI,wBAAwB,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC,EAAE,CAAC;gBACnE,MAAM,gBAAgB,GAAG,GAAG,QAAQ,IAAI,SAAS,CAAC,CAAC,CAAC,IAAI,YAAY,EAAE,CAAA;gBAEtE,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;gBAE1C,MAAM,eAAe,GAAG,GAAG,IAAI,GAAG,QAAQ,IAAI,SAAS,CAAC,CAAC,CAAC,EAAE,CAAA;gBAC5D,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;gBAExC,SAAS,IAAI,CAAC,CAAA;YACf,CAAC;iBAAM,CAAC;gBACP,MAAM,gBAAgB,GAAG,GAAG,QAAQ,IAAI,YAAY,EAAE,CAAA;gBAEtD,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;gBAE1C,MAAM,eAAe,GAAG,GAAG,IAAI,GAAG,QAAQ,EAAE,CAAA;gBAC5C,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;gBAExC,SAAS,IAAI,CAAC,CAAA;YACf,CAAC;QACF,CAAC;aAAM,CAAC;YACP,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAC9B,CAAC;IACF,CAAC;IAED,OAAO,EAAE,mBAAmB,EAAE,kBAAkB,EAAE,CAAA;AACnD,CAAC;AAED,MAAM,UAAU,4BAA4B,CAAC,UAAkB;IAC9D,MAAM,mBAAmB,GAAG,UAAU,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAClE,MAAM,oBAAoB,GAAG,UAAU,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEnE,IAAI,gBAAwB,CAAA;IAE5B,IAAI,mBAAmB,IAAI,EAAE,IAAI,oBAAoB,IAAI,EAAE,EAAE,CAAC;QAC7D,gBAAgB,GAAG,GAAG,mBAAmB,IAAI,oBAAoB,EAAE,CAAA;IACpE,CAAC;SAAM,IAAI,mBAAmB,IAAI,EAAE,IAAI,mBAAmB,GAAG,EAAE,IAAI,CAAC,IAAI,oBAAoB,GAAG,EAAE,EAAE,CAAC;QACpG,gBAAgB,GAAG,GAAG,mBAAmB,OAAO,oBAAoB,EAAE,CAAA;IACvE,CAAC;SAAM,CAAC;QACP,gBAAgB,GAAG,UAAU,CAAA;IAC9B,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,MAAM,UAAU,8BAA8B,CAAC,YAAoB;IAClE,MAAM,mBAAmB,GAAG,QAAQ,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAClE,MAAM,oBAAoB,GAAG,QAAQ,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEnE,IAAI,gBAAwB,CAAA;IAE5B,MAAM,uBAAuB,GAAG,mBAAmB,GAAG,EAAE,CAAA;IACxD,MAAM,iBAAiB,GAAI,mBAAmB,GAAG,EAAE,IAAI,CAAC,IAAI,oBAAoB,IAAI,CAAC,CAAA;IAErF,IAAI,CAAC,uBAAuB,IAAI,CAAC,iBAAiB,EAAE,CAAC;QACpD,IAAI,oBAAoB,IAAI,CAAC,EAAE,CAAC;YAC/B,gBAAgB,GAAG,GAAG,mBAAmB,IAAI,oBAAoB,GAAG,CAAA;QACrE,CAAC;aAAM,CAAC;YACP,gBAAgB,GAAG,GAAG,mBAAmB,WAAW,CAAA;QACrD,CAAC;IACF,CAAC;SAAM,CAAC;QACP,gBAAgB,GAAG,YAAY,CAAA;IAChC,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,MAAM,UAAU,6BAA6B,CAAC,IAAY;IACzD,OAAO,+BAA+B,CAAC,IAAI,EAAE,6BAA6B,CAAC,CAAA;AAC5E,CAAC;AAED,MAAM,CAAC,MAAM,6BAA6B,GAA2B;IACpE,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IAET,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;CACR,CAAA"}
|
|
@@ -100,6 +100,8 @@ export declare function loadPackagesAndGetPaths(modelName: WhisperModelName | un
|
|
|
100
100
|
export declare function normalizeWhisperModelName(modelName: WhisperModelName, languageCode: string | undefined): WhisperModelName;
|
|
101
101
|
export declare function isMultilingualModel(modelName: WhisperModelName): boolean;
|
|
102
102
|
export declare function isEnglishOnlyModel(modelName: WhisperModelName): boolean;
|
|
103
|
+
export declare function getDefaultEncoderProvidersForModel(modelName: WhisperModelName): OnnxExecutionProvider[];
|
|
104
|
+
export declare function getDefaultDecoderProvidersForModel(modelName: WhisperModelName): OnnxExecutionProvider[];
|
|
103
105
|
export type WhisperTokenData = {
|
|
104
106
|
id: number;
|
|
105
107
|
text: string;
|
|
@@ -124,6 +126,7 @@ export interface WhisperOptions {
|
|
|
124
126
|
decodeTimestampTokens?: boolean;
|
|
125
127
|
endTokenThreshold?: number;
|
|
126
128
|
includeEndTokenInCandidates?: boolean;
|
|
129
|
+
useOptimizedAlignmentHeads?: boolean;
|
|
127
130
|
encoderProvider?: OnnxExecutionProvider;
|
|
128
131
|
decoderProvider?: OnnxExecutionProvider;
|
|
129
132
|
seed?: number;
|
|
@@ -132,6 +135,8 @@ export declare const defaultWhisperOptions: WhisperOptions;
|
|
|
132
135
|
export interface WhisperAlignmentOptions {
|
|
133
136
|
model?: WhisperModelName;
|
|
134
137
|
endTokenThreshold?: number;
|
|
138
|
+
maxTokensPerPart?: number;
|
|
139
|
+
useOptimizedAlignmentHeads?: boolean;
|
|
135
140
|
encoderProvider?: OnnxExecutionProvider;
|
|
136
141
|
decoderProvider?: OnnxExecutionProvider;
|
|
137
142
|
}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
import { Logger } from '../utilities/Logger.js';
|
|
2
2
|
import { computeMelSpectogramUsingFilterbanks } from '../dsp/MelSpectogram.js';
|
|
3
|
-
import { clip,
|
|
3
|
+
import { clip, getIntegerRange, splitFloat32Array, yieldToEventLoop } from '../utilities/Utilities.js';
|
|
4
4
|
import { indexOfMax, logOfVector, logSumExp, meanOfVector, softmax, stdDeviationOfVector } from '../math/VectorMath.js';
|
|
5
5
|
import { alignDTWWindowed } from '../alignment/DTWSequenceAlignmentWindowed.js';
|
|
6
6
|
import { extendDeep } from '../utilities/ObjectUtilities.js';
|
|
@@ -15,8 +15,9 @@ import { detectSpeechLanguageByParts } from '../api/SpeechLanguageDetection.js';
|
|
|
15
15
|
import { isPunctuation, isWhitespace, isWord, splitToSentences, splitToWords } from '../nlp/Segmentation.js';
|
|
16
16
|
import { medianOf5Filter } from '../math/MedianFilter.js';
|
|
17
17
|
import { getDeflateCompressionMetricsForString } from '../utilities/Compression.js';
|
|
18
|
-
import { getOnnxSessionOptions, makeOnnxLikeFloat32Tensor } from '../utilities/OnnxUtilities.js';
|
|
18
|
+
import { dmlProviderAvailable, getOnnxSessionOptions, makeOnnxLikeFloat32Tensor } from '../utilities/OnnxUtilities.js';
|
|
19
19
|
import { murmurHash3_int32Input } from '../utilities/Hashing.js';
|
|
20
|
+
import { containsInvalidCodepoint, getTokenRepetitionScore } from '../utilities/StringUtilities.js';
|
|
20
21
|
export async function recognize(sourceRawAudio, modelName, modelDir, task, sourceLanguage, options) {
|
|
21
22
|
options = extendDeep(defaultWhisperOptions, options);
|
|
22
23
|
if (sourceRawAudio.sampleRate != 16000) {
|
|
@@ -256,7 +257,7 @@ export class Whisper {
|
|
|
256
257
|
partCrossAttentionQKs = partCrossAttentionQKs.slice(initialTokens.length);
|
|
257
258
|
// Find alignment path
|
|
258
259
|
let alignmentHeads = undefined;
|
|
259
|
-
if (
|
|
260
|
+
if (options.useOptimizedAlignmentHeads) {
|
|
260
261
|
alignmentHeads = this.alignmentHeadIndexes;
|
|
261
262
|
}
|
|
262
263
|
const alignmentPath = await this.findAlignmentPathFromQKs(partCrossAttentionQKs, partTokens, 0, segmentFrameCount, alignmentHeads);
|
|
@@ -337,12 +338,15 @@ export class Whisper {
|
|
|
337
338
|
topCandidateCount: 1,
|
|
338
339
|
punctuationThreshold: Infinity,
|
|
339
340
|
autoPromptParts: false,
|
|
340
|
-
maxTokensPerPart:
|
|
341
|
+
maxTokensPerPart: whisperAlignmentOptions.maxTokensPerPart,
|
|
341
342
|
suppressRepetition: false,
|
|
342
343
|
repetitionThreshold: Infinity,
|
|
343
344
|
decodeTimestampTokens: true,
|
|
344
345
|
endTokenThreshold: whisperAlignmentOptions.endTokenThreshold,
|
|
345
346
|
includeEndTokenInCandidates: false,
|
|
347
|
+
useOptimizedAlignmentHeads: whisperAlignmentOptions.useOptimizedAlignmentHeads,
|
|
348
|
+
encoderProvider: whisperAlignmentOptions.encoderProvider,
|
|
349
|
+
decoderProvider: whisperAlignmentOptions.decoderProvider,
|
|
346
350
|
seed: undefined,
|
|
347
351
|
};
|
|
348
352
|
// Recognize
|
|
@@ -731,7 +735,8 @@ export class Whisper {
|
|
|
731
735
|
const audioSamples = rawAudio.audioChannels[0];
|
|
732
736
|
const sampleRate = rawAudio.sampleRate;
|
|
733
737
|
const fftOrder = 400;
|
|
734
|
-
const
|
|
738
|
+
const fftWindowSize = 400;
|
|
739
|
+
const fftHopLength = 160;
|
|
735
740
|
const filterbankCount = this.filterbankCount;
|
|
736
741
|
const filterbanks = this.filterbanks;
|
|
737
742
|
const maxAudioSamples = sampleRate * 30;
|
|
@@ -748,7 +753,7 @@ export class Whisper {
|
|
|
748
753
|
const paddedAudioSamples = new Float32Array(maxAudioSamples);
|
|
749
754
|
paddedAudioSamples.set(audioSamples, 0);
|
|
750
755
|
const rawAudioPart = { audioChannels: [paddedAudioSamples], sampleRate };
|
|
751
|
-
const { melSpectogram } = await computeMelSpectogramUsingFilterbanks(rawAudioPart, fftOrder,
|
|
756
|
+
const { melSpectogram } = await computeMelSpectogramUsingFilterbanks(rawAudioPart, fftOrder, fftWindowSize, fftHopLength, filterbanks);
|
|
752
757
|
await logger.startAsync('Normalize mel spectogram');
|
|
753
758
|
const logMelSpectogram = melSpectogram.map(spectrum => spectrum.map(mel => Math.log10(Math.max(mel, 1e-10))));
|
|
754
759
|
// Find maximum log mel value in the spectrum
|
|
@@ -1286,17 +1291,17 @@ export function isMultilingualModel(modelName) {
|
|
|
1286
1291
|
export function isEnglishOnlyModel(modelName) {
|
|
1287
1292
|
return modelName.endsWith('.en');
|
|
1288
1293
|
}
|
|
1289
|
-
function getDefaultEncoderProvidersForModel(modelName) {
|
|
1290
|
-
if (
|
|
1294
|
+
export function getDefaultEncoderProvidersForModel(modelName) {
|
|
1295
|
+
if (dmlProviderAvailable()) {
|
|
1291
1296
|
return ['dml', 'cpu'];
|
|
1292
1297
|
}
|
|
1293
1298
|
else {
|
|
1294
1299
|
return [];
|
|
1295
1300
|
}
|
|
1296
1301
|
}
|
|
1297
|
-
function getDefaultDecoderProvidersForModel(modelName) {
|
|
1302
|
+
export function getDefaultDecoderProvidersForModel(modelName) {
|
|
1298
1303
|
if (modelName.startsWith('small') || modelName.startsWith('medium') || modelName.startsWith('large')) {
|
|
1299
|
-
if (
|
|
1304
|
+
if (dmlProviderAvailable()) {
|
|
1300
1305
|
return ['dml', 'cpu'];
|
|
1301
1306
|
}
|
|
1302
1307
|
else {
|
|
@@ -1662,6 +1667,7 @@ export const defaultWhisperOptions = {
|
|
|
1662
1667
|
decodeTimestampTokens: true,
|
|
1663
1668
|
endTokenThreshold: 0.9,
|
|
1664
1669
|
includeEndTokenInCandidates: true,
|
|
1670
|
+
useOptimizedAlignmentHeads: true,
|
|
1665
1671
|
encoderProvider: undefined,
|
|
1666
1672
|
decoderProvider: undefined,
|
|
1667
1673
|
seed: undefined,
|
|
@@ -1669,8 +1675,10 @@ export const defaultWhisperOptions = {
|
|
|
1669
1675
|
export const defaultWhisperAlignmentOptions = {
|
|
1670
1676
|
model: undefined,
|
|
1671
1677
|
endTokenThreshold: 0.9,
|
|
1678
|
+
maxTokensPerPart: 250,
|
|
1679
|
+
useOptimizedAlignmentHeads: true,
|
|
1672
1680
|
encoderProvider: undefined,
|
|
1673
|
-
decoderProvider: undefined
|
|
1681
|
+
decoderProvider: undefined,
|
|
1674
1682
|
};
|
|
1675
1683
|
export const defaultWhisperLanguageDetectionOptions = {
|
|
1676
1684
|
model: undefined,
|