Chinese · Mandarin
Chinese text to speech that sounds like a person
Paste Mandarin text and get natural speech in seconds. Tune pace, pitch and emotion, then download MP3 or WAV.
How VoiceAir handles Mandarin
VoiceAir runs on Doubao Seed TTS 2.0, a model that reads the sentence before speaking it. That means it resolves polyphonic characters from context, reads numbers and units correctly, and keeps a natural rhythm across long sentences instead of pausing after every clause.
You can adjust speech rate from -50% to +100%, pitch from -12 to +12 semitones, and loudness from -50% to +100%. Nine emotions (neutral, happy, sad, angry, surprised, fear, hate, excited, coldness) are available with an intensity slider, although not every voice supports emotion control.
Output formats and limits
Audio can be exported as MP3, WAV or OGG Opus at 8, 16, 22.05, 24, 32, 44.1 or 48 kHz. One request handles up to 1,000 characters, which is roughly two minutes of speech; longer scripts should be generated section by section and joined in an editor.
Files are generated on demand and never stored on the server. The free tier includes 3,000 characters per month for personal use, and paid plans add a commercial licence.
Who uses it
Cross-border sellers record Chinese listings and then produce English or Japanese versions from the same script. Product teams narrate demo videos without booking a studio. Teachers build listening material at reduced speed for learners.
Frequently asked questions
Does it support Cantonese?
Not yet. VoiceAir focuses on Mandarin, English, Japanese and Korean.
Can I use the audio commercially?
Commercial use is included on every paid plan. The free tier is limited to personal trials.
How long can one generation be?
Up to 1,000 characters per request, which is roughly two minutes of speech.
Start with 3,000 free characters
No credit card required. Translate a Chinese script and download the English, Japanese or Korean voice over in minutes.