VoiceStudio is the open-source, fully-local ElevenLabs alternative — voice cloning, voice design, video dubbing, dictation, transcription & audiobook creation in 646 languages.
Python aiaudiobookcudadubbingelevenlabs-alternativehuggingfacelocal-firstmlxomnivoice-studiospeech-to-texttext-to-speechtranscriptiontranslatettsvoice-aivoice-cloningvoice-generationvoicestudioworkflow
import { describe, it, expect } from 'vitest';
import { parseSrt, importToText } from './importStory';
const SRT = `1
00:00:01,000 --> 00:00:03,000
Hello there.
2
00:00:03,500 --> 00:00:05,000
How are you?
`;
describe('parseSrt', () => {
it('strips indices + timestamps, one cue per line', () => {
expect(parseSrt(SRT)).toBe('Hello there.\nHow are you?');
});
it('joins multi-line cues with a space', () => {
const s = '1\n00:00:01,000 --> 00:00:02,000\nLine one\nline two\n';
expect(parseSrt(s)).toBe('Line one line two');
});
it('empty → empty', () => {
expect(parseSrt('')).toBe('');
});
it('keeps a cue whose dialogue is only a number', () => {
const s =
'1\n00:00:01,000 --> 00:00:02,000\nCount down with me.\n\n' +
'2\n00:00:02,000 --> 00:00:03,000\n3\n\n' +
'3\n00:00:03,000 --> 00:00:04,000\n2\n1\n';
expect(parseSrt(s)).toBe('Count down with me.\n3\n2 1');
});
it('keeps a numeric cue in a compact file whose next cue has no index', () => {
const s =
'00:00:01,000 --> 00:00:02,000\nCount down with me.\n' +
'00:00:02,000 --> 00:00:03,000\n3\n' +
'00:00:03,000 --> 00:00:04,000\n2\n';
expect(parseSrt(s)).toBe('Count down with me. 3 2');
});
it('keeps numeric dialogue at the end of a multi-line cue in a compact file', () => {
const s =
'00:00:01,000 --> 00:00:02,000\nCount down with me.\n3\n' +
'00:00:02,000 --> 00:00:03,000\nNext\n';
expect(parseSrt(s)).toBe('Count down with me. 3 Next');
});
it('preserves dialogue equal to the next index when cues are unnumbered', () => {
expect(
parseSrt('00:00:01,000 --> 00:00:02,000\nCount\n2\n00:00:02,000 --> 00:00:03,000\nNext'),
).toBe('Count 2 Next');
});
it('still drops the indices of a compact file', () => {
const s =
'1\n00:00:01,000 --> 00:00:02,000\nCount\n' +
'2\n00:00:02,000 --> 00:00:03,000\n3\n' +
'3\n00:00:03,000 --> 00:00:04,000\n2\n';
expect(parseSrt(s)).toBe('Count 3 2');
});
it('drops karaoke tags and alignment overrides from spoken lines', () => {
const s =
'1\n00:00:01,000 --> 00:00:02,000\n{\\an8}{\\i1}Hello{\\i0}\n\n' +
'2\n00:00:02,000 --> 00:00:03,000\nhey<00:00:00.480><c> everyone</c>\n';
expect(parseSrt(s)).toBe('Hello\nhey everyone');
});
it('keeps a literal less-than that is not a tag', () => {
expect(parseSrt('1\n00:00:01,000 --> 00:00:02,000\nI <3 you\n')).toBe('I <3 you');
});
it('keeps angle-bracketed dialogue that is not a player tag', () => {
for (const line of ['2 < 3 and 4 > 1', '<laughter> okay', 'if a<b and c>d']) {
expect(parseSrt(`1\n00:00:01,000 --> 00:00:02,000\n${line}\n`)).toBe(line);
}
});
it('drops font, voice and bold tags but keeps CJK and RTL words', () => {
const words = '\u4f60\u597d \u05e9\u05dc\u05d5\u05dd';
const s = `1\n00:00:01,000 --> 00:00:02,000\n<font color="#ff0">{\\an8}<b>${words}</b></font> <v Roger>ok</v>\n`;
expect(parseSrt(s)).toBe(`${words} ok`);
});
it('reads a SubRip line break as a word gap', () => {
expect(parseSrt('1\n00:00:01,000 --> 00:00:02,000\nHello<br>big<BR />world\n')).toBe(
'Hello big world',
);
});
it('scans unclosed markup prefixes in linear time', () => {
const started = Date.now();
parseSrt(`1\n00:00:01,000 --> 00:00:02,000\n${'<i'.repeat(50000)}${'{\\'.repeat(50000)}\n`);
parseSrt(`WEBVTT\n\n00:00:01.000 --> 00:00:02.000\n${'<'.repeat(50000)}\n`);
expect(Date.now() - started).toBeLessThan(1000);
});
it('does not speak a WebVTT file header', () => {
const vtt =
'WEBVTT\n\n00:00:01.000 --> 00:00:02.000\nHello\n\n00:00:02.000 --> 00:00:03.000\nWorld\n';
expect(parseSrt(vtt)).toBe('Hello\nWorld');
});
it('skips WebVTT timings with three-digit hours instead of speaking them', () => {
expect(parseSrt('WEBVTT\n\n100:00:00.000 --> 100:00:02.000\nLong recording\n')).toBe(
'Long recording',
);
});
it('drops NOTE, STYLE and REGION blocks and cue identifiers', () => {
const vtt =
'WEBVTT\n\nNOTE made by a translator\n\n' +
'intro\n00:00:01.000 --> 00:00:02.500 align:start\nHola\n\n' +
'STYLE\n::cue { color: red }\n\n' +
'REGION\nid:fred width:40%\n\n' +
'cue-2\n00:00:03.000 --> 00:00:04.000\nQue tal\n';
expect(parseSrt(vtt)).toBe('Hola\nQue tal');
});
it('does not speak arrow-bearing WebVTT metadata without a timing line', () => {
const vtt =
'WEBVTT\n\nNOTE\nTransition A --> B\nprivate note\n\n' +
'STYLE\nTransition A --> B\n::cue { color: red }\n\n' +
'00:00:01.000 --> 00:00:02.000\nSpoken text\n';
expect(parseSrt(vtt)).toBe('Spoken text');
});
it('lets a timing line in position two outrank a STYLE or REGION identifier', () => {
for (const identifier of ['STYLE', 'REGION']) {
expect(parseSrt(`WEBVTT\n\n${identifier}\n00:01.000 --> 00:02.000\nSpoken text\n`)).toBe(
'Spoken text',
);
}
});
it('never treats a NOTE block as a cue even when its second line is a timing line', () => {
for (const identifier of ['NOTE', 'NOTE identifier']) {
const vtt =
`WEBVTT\n\n${identifier}\n00:00:01.000 --> 00:00:02.000\nprivate note\n\n` +
'00:00:03.000 --> 00:00:04.000\nSpoken text\n';
expect(parseSrt(vtt)).toBe('Spoken text');
}
});
it('keeps a cue whose identifier is STYLE or REGION in a mixed file', () => {
for (const identifier of ['intro', 'STYLE', 'REGION']) {
const vtt =
`WEBVTT\n\n${identifier}\n00:00:01.000 --> 00:00:02.000\nHello world.\n\n` +
'next\n00:00:03.000 --> 00:00:04.000\nStill here.\n';
expect(parseSrt(vtt)).toBe('Hello world.\nStill here.');
}
});
it('keeps NOTE when it is the spoken dialogue', () => {
expect(
parseSrt('WEBVTT\n\n00:01.000 --> 00:02.000\nNOTE this is spoken\nSTYLE\nREGION\n'),
).toBe('NOTE this is spoken STYLE REGION');
});
it('unescapes WebVTT character references after dropping tags', () => {
expect(parseSrt('WEBVTT\n\n00:00:01.000 --> 00:00:02.000\nTom & Jerry\n')).toBe(
'Tom & Jerry',
);
expect(
parseSrt('WEBVTT\n\n00:00:01.000 --> 00:00:02.000\n<i>literal</i> <i>real</i>\n'),
).toBe('<i>literal</i> real');
});
it('decodes other HTML references and directional marks in WebVTT speech', () => {
expect(parseSrt('WEBVTT\n\n00:01.000 --> 00:02.000\nFrançais ‎gauche‏\n')).toBe(
'Français \u200egauche\u200f',
);
});
it('still keeps a SubRip entity as written', () => {
expect(parseSrt('1\n00:00:01,000 --> 00:00:02,000\nTom & Jerry\n')).toBe('Tom & Jerry');
});
it('reads a caption file that uses only CR line endings', () => {
expect(
parseSrt(
'1\r00:00:01,000 --> 00:00:02,000\rHello\r\r2\r00:00:02,000 --> 00:00:03,000\rWorld',
),
).toBe('Hello\nWorld');
});
});
describe('importToText', () => {
it('routes .srt through parseSrt', () => {
expect(importToText('subs.srt', SRT)).toBe('Hello there.\nHow are you?');
});
it('passes .txt through unchanged', () => {
expect(importToText('story.txt', 'Once upon a time.')).toBe('Once upon a time.');
});
it('routes .vtt through parseSrt so timestamps are not spoken', () => {
const vtt =
'WEBVTT\n\n00:00:00.160 --> 00:00:02.310 align:start position:0%\nhey<00:00:00.480><c> everyone</c>\n';
expect(importToText('captions.vtt', vtt)).toBe('hey everyone');
expect(importToText('captions.VTT', vtt)).toBe('hey everyone');
});
});