@jackdbd/eleventy-plugin-text-to-speech/lib/synthesis/elevenlabs-text-to-speech.js

Eleventy plugin that uses text-to-speech to generate audio assets for your website, then injects audio players in your HTML.

import { Readable } from 'node:stream';
import defDebug from 'debug';
import { z } from 'zod';
import { elevenlabs } from '@jackdbd/zod-schemas';
import { DEBUG_PREFIX } from '../constants.js';
import { mediaType } from '../media-type.js';
import { validatedDataOrThrow } from '../validation.js';
const debug = defDebug(`${DEBUG_PREFIX}:elevenlabs-text-to-speech`);
export const audioExtension = (outputFormat) => {
    switch (outputFormat) {
        case 'mp3_44100_64':
        case 'mp3_44100_96':
        case 'mp3_44100_128':
        case 'mp3_44100_192':
            return 'mp3';
        // https://en.wikipedia.org/wiki/Pulse-code_modulation
        case 'pcm_16000':
        case 'pcm_22050':
        case 'pcm_24000':
        case 'pcm_44100':
            return 'pcm';
        // https://en.wikipedia.org/wiki/%CE%9C-law_algorithm
        // https://en.wikipedia.org/wiki/FLAC
        case 'ulaw_8000':
            return 'flac'; // or mulaw?
        default:
            return 'mp3';
    }
};
export const DEFAULT_VOICE_SETTINGS = {
    similarity_boost: 0.5,
    stability: 0.5,
    style: 1,
    use_speaker_boost: true
};
export const synthesis_config = z
    .object({
    /**
     * Models supported by the ElevenLabs Text-to-Speech API.
     *
     * @see https://elevenlabs.io/docs/api-reference/get-models
     */
    modelId: elevenlabs.model_id,
    outputFormat: elevenlabs.output_format,
    /**
     * Voices supported by the ElevenLabs Text-to-Speech API.
     *
     * @see https://elevenlabs.io/docs/api-reference/get-voices
     */
    voiceId: elevenlabs.voice_id,
    voiceSettings: z.any().default(DEFAULT_VOICE_SETTINGS)
})
    .describe('ElevenLabs synthesis config');
export const synthesize_config = z
    .object({
    /**
     * Text to synthesize into speech using the ElevenLabs Text-to-Speech API.
     *
     * @remarks
     * Character limit for the ElevenLabs Text-to-Speech API:
     * - 2500 for unsubscribed users
     * - 5000 for subscribed users
     * @see [elevenlabs.io - Pricing](https://elevenlabs.io/pricing)
     */
    text: elevenlabs.text
})
    .describe('ElevenLabs synthesize config');
/**
 * Synthesizes text into speech using the ElevenLabs Text-to-Speech API.
 *
 * @public
 */
export const synthesize = async (apiKey, cfg, text) => {
    const { modelId, outputFormat, voiceId, voiceSettings } = cfg;
    const reqBody = {
        model_id: modelId,
        output_format: outputFormat,
        // pronunciation_dictionary_locators: [
        //   { pronunciation_dictionary_id: '<string>', version_id: '<string>' }
        // ],
        text,
        voice_settings: voiceSettings
    };
    const ext = `.${audioExtension(outputFormat)}`;
    const reqInit = {
        method: 'POST',
        headers: {
            Accept: mediaType(ext).value,
            'Content-Type': 'application/json',
            'xi-api-key': apiKey
        },
        body: JSON.stringify(reqBody)
    };
    const apiEndpoint = 'https://api.elevenlabs.io/v1/text-to-speech';
    let response;
    try {
        response = await fetch(`${apiEndpoint}/${voiceId}`, reqInit);
        // eslint-disable-next-line @typescript-eslint/no-explicit-any
    }
    catch (err) {
        return {
            error: new Error(`Could not synthesize text: failed to fetch: ${err.message}`)
        };
    }
    if (!response.body) {
        return {
            error: new Error(`Could not synthesize text: no body`)
        };
    }
    if (response.status !== 200) {
        return {
            error: new Error(`Could not synthesize text: ${response.statusText}`)
        };
    }
    const readable = Readable.from(response.body);
    readable.on('error', (error) => {
        debug(`readable stream errored`);
        return { error };
    });
    readable.on('end', () => {
        debug(`readable stream ended, no more data to be consumed from`);
    });
    readable.on('close', async () => {
        debug(`readable stream closed, all underlying resources freed`);
    });
    return { value: readable };
};
export const auth_options = z.object({
    apiKey: elevenlabs.api_key.optional()
});
export const client_config = auth_options.merge(synthesis_config);
/**
 * Client for the ElevenLabs Text-to-Speech API.
 */
export const defClient = (config) => {
    const data = validatedDataOrThrow(config, client_config);
    const { modelId, outputFormat, voiceId } = data;
    const apiKey = data.apiKey || process.env.ELEVENLABS_API_KEY || '';
    const voiceSettings = data.voiceSettings || DEFAULT_VOICE_SETTINGS;
    const extension = audioExtension(data.outputFormat);
    debug(`text will be synthesized into [hash].${extension} using model ${modelId}, voice ${voiceId}, voice settings %O`, voiceSettings);
    const synthesizeWithApiKeyAndSynthesisConfig = synthesize.bind(null, apiKey, {
        modelId,
        outputFormat,
        voiceId,
        voiceSettings
    });
    return {
        config: { modelId, outputFormat, voiceId },
        extension,
        synthesize: synthesizeWithApiKeyAndSynthesisConfig
    };
};
//# sourceMappingURL=elevenlabs-text-to-speech.js.map