Files
tts/scripts/generate_hume_speech.js

66 lines
1.8 KiB
JavaScript

import fs from 'fs';
import path from 'path';
import { program } from 'commander';
program
.requiredOption('-t, --text <text>', 'Text to convert to speech')
.requiredOption('-o, --output <path>', 'Output file path (e.g., output.mp3)')
.option('-v, --voice <voice>', 'Voice ID to use', '9e1f9e4f-691a-4bb0-b87c-e306a4c838ef')
.parse(process.argv);
const options = program.opts();
async function main() {
const apiKey = process.env.HUME_API_KEY;
if (!apiKey) {
console.error('Error: HUME_API_KEY environment variable must be set.');
process.exit(1);
}
try {
console.log(`Generating speech via Hume AI for: "${options.text.substring(0, 50)}..."`);
const response = await fetch('https://api.hume.ai/v0/tts', {
method: 'POST',
headers: {
'X-Hume-Api-Key': apiKey,
'Content-Type': 'application/json'
},
body: JSON.stringify({
utterances: [
{
text: options.text,
voice: {
id: options.voice
}
}
]
})
});
if (!response.ok) {
const errorText = await response.text();
throw new Error(`Hume API error (${response.status}): ${errorText}`);
}
const json = await response.json();
if (!json.generations || json.generations.length === 0) {
throw new Error('Hume API returned no audio generations.');
}
const base64Audio = json.generations[0].audio;
const buffer = Buffer.from(base64Audio, 'base64');
await fs.promises.writeFile(options.output, buffer);
const absolutePath = path.resolve(options.output);
console.log(`Audio saved: ${absolutePath}`);
console.log(`MEDIA: ${absolutePath}`);
} catch (error) {
console.error('Error generating speech with Hume:', error);
process.exit(1);
}
}
main();