Addis AI
All languages

Sidaamu Afoo

Sidama voices and speech-to-text

Addis AI speaks and understands Sidama (Sidaamu Afoo). Turn Sidama text into natural speech with 5 voices, and turn Sidama audio into text with Addis Scribe, from uploaded files or live audio.

What you can use today

Speech to text (Addis Scribe)
Uploads and live audio
Speaker labels
Yes
Captions
Word timestamps, SRT and VTT
Voices
5 voices
Streaming text to speech
Yes
Chat and translation
Not available

Sidama speech-to-text accuracy

33.6% word error rate in live mode, against 76% for Google Gemini 3 Flash (preview) (whole file). Lower is better.

Measured on WAXAL, 150 clips. Ethio-ASR, an open whole-file model, is slightly more accurate (33.0%).

Method, test sets and full results

Sidama voices

Each preview is a clip generated by the voice.

  • TigistDefault

    A clear, confident female voice with a broadcast delivery.

    sid-tigist

  • Desta

    A clear, steady male voice with a calm news delivery.

    sid-desta

  • Bereket

    A full, even male voice suited to narration.

    sid-bereket

  • Selam

    A warm female voice with an engaging presenter tone.

    sid-selam

  • Alemu

    A deep, measured male voice with a formal pace.

    sid-alemu

Questions about Sidama

Does Addis AI have Sidama text-to-speech voices?
Yes. 5 Sidama voices are available through the Addis AI API, and Tigist is the default voice. You can play a preview of each voice on this page.
Can Addis Scribe transcribe Sidama?
Yes. Addis Scribe turns Sidama speech into text from uploaded files and live audio (language code "sid"), with word timestamps and SRT or VTT captions. It can also label who is speaking.
Is real-time Sidama voice available?
Yes. Sidama works with streaming text-to-speech and with the real-time voice API.