addisai/addis-scribe-tigrinya-streaming

Addis Scribe Streaming: Tigrinya

Realtime Tigrinya speech recognition by Addis AI. Text appears while people speak and is final shortly after they stop. Trained by Addis AI on Tigrinya speech. The same model runs Addis Scribe's language=ti in the Addis AI API.

Use it

import nemo.collections.asr as nemo_asr

model = nemo_asr.models.ASRModel.restore_from("addis-scribe-tigrinya-streaming.nemo")
model.set_inference_prompt("am-ET")          # Tigrinya uses prompt index 49 (the am-ET slot)
print(model.transcribe(["speech_16k.wav"])[0])

For live use, run it as a cache-aware streaming model with 0.32 s or 1.12 s chunks, the same way as Addis Scribe Streaming (Amharic) and its stream_infer.py, with prompt index 49. Requires NeMo 3.0. Or call it through the Addis AI API with language=ti, including speaker labels, word timestamps and captions.

Model

Architecture	Cache-aware FastConformer encoder, hybrid RNNT + CTC, 0.6B parameters
Decoding	Greedy RNNT; prompt index 49
Chunk sizes	0.32 s or 1.12 s
Input / output	16 kHz mono audio / Tigrinya text in Ethiopic script, no punctuation
Files	addis-scribe-tigrinya-streaming.nemo, SHA256SUMS

Limits

• No punctuation.
• Tigrinya spelling varies (for example ፀ/ጸ and ሰ/ሠ); some measured errors are spelling differences from the reference rather than mishearing.
• Tested on Tigrinya only.

License

OpenMDW 1.1, the same as the base model's terms allow. Made by Addis AI.