G2

Aryan K.

AI Engineer

Small-Business (50 or fewer emp.)

10/19/2026

"Low-Latency Speech and Great Voice Cloning, with Occasional Delays"

4.5/5

What do you like best about smallest.ai?

I use smallest.ai for speech-to-text and text-to-speech, and I find the latency of both good. I especially like the low latency of speech-to-text, the text-to-speech quality, and the range of languages and human-sounding voices. Voice cloning is fast and works really well with around 5 to 15 seconds of audio. The range of voices and voice cloning helped me simulate different personas for role-play agents. I can support Indian languages like Malayalam and Kannada, and work with clients whose recipients speak those languages. The setup was very smooth; I used Claude to search the smallest.ai documentation and get the streaming API connections working. Review collected by and hosted on G2.com.

What do you dislike about smallest.ai?

Sometimes the text-to-speech latency increases a bit, though it doesn’t happen every time. I think smallest.ai should allow more voice samples for cloning, because 15 seconds of audio may not capture all the expressions and pauses, and sometimes the voice is unable to clone. Review collected by and hosted on G2.com.

What problems is smallest.ai solving and how is that benefiting you?

I use smallest.ai’s speech-to-text and text-to-speech to build my own inbound and outbound calling infrastructure. Low latency, human-sounding voices in many languages, and voice cloning from 5–15 seconds of audio help me simulate personas and serve clients whose recipients speak languages like Malayalam and Kannada. Review collected by and hosted on G2.com.