Tamil AI Voice

Kani — free Malaysian Tamil text to speech

Kani is a free female Malaysian Tamil AI voice — a Microsoft neural voice. Type your text below to hear it read aloud, then download the audio. No account needed.

Female Malaysian accent Global voice Free

Make Kani say your words

Voice details

Language
Tamil
Accent
Malaysian
Gender
Female
Type
Global · Microsoft neural
Download
MP3
Speed
Natural pace
Voice ID
ta-MY-KaniNeural
Price
Free

How Kani sounds — measured

Median pitch

254 Hz

The highest of 4 female Tamil voices

Intonation range

10.5 semitones

Close to typical for its group

Tone (spectral centroid)

1,715 Hz

Balanced tone for its group

Speaking rate

13.9 characters/s

The 2nd fastest of 8 Tamil voices on the same passage

Pitch across the sample passage dashed line = Kani’s median

How long your audio will be at 13.9 characters/s

1,000 characters
≈ 1 min 12 s
5,000 characters (free limit per render)
≈ 6 min

Measured by SpeakScribe, October 2026 · How we measure · Every Tamil voice side by side

What Kani fits best

  • Tutorials, kids' content and friendly explainers

    At 254 Hz it sits above 3 of the 4 female Tamil voices, which sounds bright and approachable.

  • Fast-paced social video

    At 13.9 characters a second it fits more text into a short clip.

Use Kani in a conversation

With Multi-voice, Kani can read one character while Valluvar (a male Tamil voice) reads another — the whole dialogue comes back as one MP3 with even volume between speakers. A free account can make 3 two-speaker dialogues a month; paid plans allow up to 4 speakers on Starter.

Try a dialogue with Kani →

Formats, limits and help

Kani: quick answers

Measured from its audio, Kani has a median pitch 254 Hz (the highest of 4 female Tamil voices); intonation range 10.5 semitones (close to typical for its group); tone (spectral centroid) 1,715 Hz (balanced tone for its group).
About 1 min 12 s. Kani reads about 13.9 characters a second, spaces included (measured on its sample passage), so the free limit of 5,000 characters per render is about 6 minutes of audio. Numbers, names and long pauses make it longer.