AudioModels class abstract final

Verified official models accepted by audio task options.

Properties

hashCode → int
The hash code for this object.
no setterinherited
runtimeType → Type
A representation of the runtime type of the object.
no setterinherited

Methods

noSuchMethod(Invocation invocation) → dynamic
Invoked when a nonexistent method or property is accessed.
inherited
toString() → String
A string representation of this object.
inherited

Operators

operator ==(Object other) → bool
The equality operator.
inherited

Constants

byName → const Map<String, DownloadAsset>
Every model above by the name an app lists to bundle it, under hooks.user_defines.mediapipe_audio.models in pubspec.yaml.
kokoro → const DownloadAsset
hexgrad's Kokoro 82M voices for Text to Speech, as Google's LiteRT community converted them (int8, with an FP16 vocoder for the GPU delegate; English, Spanish, French, Italian, Portuguese, Hindi, Chinese and Japanese at 24 kHz): 295 MB, Apache 2.0.
parakeetTdtV3 → const DownloadAsset
NVIDIA's Parakeet TDT 0.6B v3 speech recognizer for the Audio Transcriber, as Google's LiteRT community converted it (int8, 5 s stateful windows, 25 European languages): 615 MB, CC BY 4.0.
phonon → const DownloadAsset
Gradium's Phonon 31M voices for Text to Speech, as Google's LiteRT community converted them (int8; five voices at 24 kHz, freya by default): 87 MB, Apache 2.0. Google's guide says Phonon speaks US English only. Its 1.2.3 runtime labels four of the voices with other locales (claire fr-FR, lukas de-DE, sofia es-ES, lucas pt-BR); for other languages, use kokoro.
pocketTtsEn → const DownloadAsset
Kyutai's PocketTTS 100M English voices for Text to Speech, as Google's LiteRT community converted them (int8, CPU only; eight US English voices at 24 kHz, alba by default): 163 MB, CC BY 4.0, so an app that ships it credits Kyutai, as in its open source notices.
yamnet → const DownloadAsset
Google's YAMNet audio classifier, version 1, as MediaPipe Studio uses it.