Property	Value
Parameters	600M
Architecture	Encoder-Decoder (Speech-to-Text)
Languages	52 (auto-detected)
Latency	~100-200ms per utterance
License	Apache 2.0
Organization	OsirisBrain

Usage

python

from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor

processor = AutoProcessor.from_pretrained("osirisbrain/osiriswhisper-stt", trust_remote_code=True)
model = AutoModelForSpeechSeq2Seq.from_pretrained("osirisbrain/osiriswhisper-stt", trust_remote_code=True)

OsirisBrain Integration

Part of the OsirisBrain Sovereign AI ecosystem:

OsirisSound TTS: Text-to-Speech (Meritaten v3.0)
OsirisWhisper STT: Speech-to-Text (this model)
OsirisCortex v6: 9B Reasoning + Vision
OsirisPtah-Coder v5: 7B Coding

API Endpoint

When running OsirisSound server:

POST http://127.0.0.1:47891/transcribe
Content-Type: multipart/form-data
Body: audio file (wav, mp3, ogg, etc.)