57 lines
1.9 KiB
TypeScript
57 lines
1.9 KiB
TypeScript
|
import { writeFile } from "node:fs/promises";
|
||
|
|
|
||
|
|
export type VoiceParams = Record<string, unknown>;
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Common contract for every TTS provider. Subclasses override either
|
||
|
|
* `getSpeech` (returning a fetch-like Response) or `getSpeechFile` (writing
|
||
|
|
* directly to disk) — the default `getSpeechFile` pipes `getSpeech` to a file.
|
||
|
|
*/
|
||
|
|
export abstract class BaseEngine {
|
||
|
|
/** Short ID used in env / commands (e.g. "azure"). */
|
||
|
|
readonly shortName: string;
|
||
|
|
/** Human-readable name shown in messages. */
|
||
|
|
readonly longName: string;
|
||
|
|
/** Output file extension without the dot (e.g. "mp3"). */
|
||
|
|
readonly fileExtension: string;
|
||
|
|
|
||
|
|
protected voices: Record<string, string | { name: string; lang: string }> = {};
|
||
|
|
|
||
|
|
constructor(shortName: string, longName: string, fileExtension: string) {
|
||
|
|
this.shortName = shortName;
|
||
|
|
this.longName = longName;
|
||
|
|
this.fileExtension = fileExtension;
|
||
|
|
}
|
||
|
|
|
||
|
|
/** Maps a user-friendly voice name to the provider's internal identifier. */
|
||
|
|
getInternalVoiceName(str: string): string {
|
||
|
|
const v = this.voices[str];
|
||
|
|
if (v == null) return str;
|
||
|
|
return typeof v === "string" ? v : v.name;
|
||
|
|
}
|
||
|
|
|
||
|
|
abstract getDefaultVoice(): string;
|
||
|
|
|
||
|
|
validateVoice(voice: string): boolean {
|
||
|
|
if (Object.keys(this.voices).length === 0) return true;
|
||
|
|
return this.voices[voice] != null;
|
||
|
|
}
|
||
|
|
|
||
|
|
/** Default implementation: subclass should override either this or getSpeechFile. */
|
||
|
|
async getSpeech(_text: string, _voice?: string, _params?: VoiceParams): Promise<Response> {
|
||
|
|
throw new Error(`${this.shortName}: getSpeech not implemented`);
|
||
|
|
}
|
||
|
|
|
||
|
|
async getSpeechFile(
|
||
|
|
text: string,
|
||
|
|
filepath: string,
|
||
|
|
voice: string = this.getDefaultVoice(),
|
||
|
|
params: VoiceParams = {},
|
||
|
|
): Promise<string> {
|
||
|
|
const data = await this.getSpeech(text, voice, params);
|
||
|
|
const buf = Buffer.from(await data.arrayBuffer());
|
||
|
|
await writeFile(filepath, buf);
|
||
|
|
return filepath;
|
||
|
|
}
|
||
|
|
}
|