About TTSLark
TTSLark is an online text to speech service for people who publish audio and video: short-form creators, narrators, language learners, and anyone who needs a clean voice track with subtitles that match it. You paste a script, pick a voice, and get an MP3 back together with SRT and VTT files timed to the same render. Every plan, including the free one, allows commercial use of the audio.
What you get
- 50+ voices in nine languages. English in American and British accents, Mandarin Chinese, Japanese, Spanish, French, Italian, Portuguese, and Hindi. Every voice has three preview clips so you can audition it before you spend a credit.
- Subtitles from the render, not from a guess. The SRT and VTT files come from the same synthesis pass as the audio, so cue boundaries land on real pauses.
- Long scripts in one take. Paid plans can submit a whole chapter or episode; the service splits it at sentence boundaries, synthesizes the pieces, and joins them into one continuous file.
- Voice blending. Mix two voices at a ratio you choose and save the result as one of your own voices.
- A script tidy pass on paid plans. An AI step adds breathing pauses to long lines, spells out numbers and units, and marks the intended reading of Chinese characters with more than one pronunciation.
How it is built
Standard voices run on speech models that TTSLark hosts itself, so free renders do not depend on a metered third-party API. Generation runs as a queue: paid jobs go first, free jobs follow after a short delay, and a progress bar shows every job as it moves. Audio and subtitle files are stored privately and handed out through expiring download links; free renders are kept for 72 hours and paid renders for 30 days, after which the files and the script text are deleted. The privacy policy describes this in detail.
What TTSLark does not do
It does not put a watermark or a spoken advertisement into free renders, and it does not hold your files longer than the retention period of your plan. Voices are named after characters, not after real people.
Who runs it
TTSLark is built and run by the TTS Lark Dev Team, a small independent team; every support message is read by the people who write the code. Pricing is deliberately simple: a free plan with five renders a month, and two subscriptions that add priority queueing, WAV output, longer retention, and the tidy pass. There is no enterprise tier and no sales call.
Contact
Write to [email protected], or use the feedback form on the contact page after signing in. Requests about a specific render should include the job id from the My Speech page.