Skip to main content
Version: Next

Type Alias: ReverbTtsModel<K>

ReverbTtsModel<K> = object

Defined in: extensions/speech/tasks/reverbTextToSpeech.ts:112

Model configuration of the Reverb Text-to-Speech pipeline.

Type Parameters​

K​

K extends PropertyKey = never

Preset voice keys.

Properties​

modelPaths​

readonly modelPaths: object

Defined in: extensions/speech/tasks/reverbTextToSpeech.ts:116

Local or remote paths to the .pte programs.

backbone​

readonly backbone: string

The backbone: prefill, step and polish, or prefill and step methods of fixed sizes.

codecDecoder​

readonly codecDecoder: string

The codec decoder: codes to audio.

codecEncoder?​

readonly optional codecEncoder: string

The codec encoder: audio to codes. Only encodeVoice loads it.


name​

readonly name: "reverb"

Defined in: extensions/speech/tasks/reverbTextToSpeech.ts:114

Discriminates this config from the other Text-to-Speech pipelines.


tokenizerPath​

readonly tokenizerPath: string

Defined in: extensions/speech/tasks/reverbTextToSpeech.ts:125

Local or remote path to tokenizer.json.


voices?​

readonly optional voices: Record<K, string>

Defined in: extensions/speech/tasks/reverbTextToSpeech.ts:127

Preset voices: name to { codes, transcript, language } JSON file.