69 lines
1.9 KiB
TypeScript
69 lines
1.9 KiB
TypeScript
import { join } from "path";
|
|
import fetch from "../utils/fetch";
|
|
import TTSModelBase from ".";
|
|
|
|
export class TTSSupertonicModel extends TTSModelBase<TTSSupertonicModel.RequestId> {
|
|
protected override cachedVoice: Map<string, Promise<Buffer>>;
|
|
constructor() {
|
|
super();
|
|
this.cachedVoice = new Map();
|
|
}
|
|
override ttsify(input: string): string {
|
|
return super.ttsify(input);
|
|
}
|
|
private async getSupertonicResponse(voiceId: TTSSupertonicModel.RequestId) {
|
|
const payload = {
|
|
text: voiceId.text,
|
|
lang: "ko",
|
|
style_id: voiceId.styleId,
|
|
};
|
|
|
|
if (!process.env.SUPERTONIC_API_URL) {
|
|
throw Error("process.env.SUPERTONIC_API_URL not set");
|
|
}
|
|
|
|
return await fetch(process.env.SUPERTONIC_API_URL, {
|
|
method: "POST",
|
|
headers: {
|
|
"Content-Type": "application/json",
|
|
},
|
|
body: JSON.stringify(payload),
|
|
});
|
|
}
|
|
async getVoiceBuffer(
|
|
voiceId: TTSSupertonicModel.RequestId,
|
|
): Promise<ArrayBuffer> {
|
|
const response: Response | undefined = (await this.getSupertonicResponse(
|
|
voiceId,
|
|
)) as Response;
|
|
if (response.ok) return await response.arrayBuffer();
|
|
|
|
throw new Error(`invalid supertonic response ${await response.text()}`);
|
|
}
|
|
public getVoicePath(id: TTSSupertonicModel.RequestId): string {
|
|
const audioFileName = TTSModelBase.hashAudioFile(id.text + id.styleId);
|
|
const audioPath = join(
|
|
TTSSupertonicModel.SupertonicAudioCachePath,
|
|
audioFileName,
|
|
);
|
|
return audioPath;
|
|
}
|
|
public createRequestId(
|
|
text: string,
|
|
styleId?: string,
|
|
): TTSSupertonicModel.RequestId {
|
|
return {
|
|
text,
|
|
styleId: styleId ?? "F1",
|
|
};
|
|
}
|
|
}
|
|
export namespace TTSSupertonicModel {
|
|
export const instance = new TTSSupertonicModel();
|
|
export type RequestId = { text: string; styleId: string };
|
|
export const SupertonicAudioCachePath = join(
|
|
TTSModelBase.AudioCachePath,
|
|
"supertonic",
|
|
);
|
|
}
|
|
export default TTSSupertonicModel;
|