Falcon-ASR
An Arabic automatic speech recognition model released by the Technology Innovation Institute (TII) of the UAE. With 1.6 billion parameters, it is designed for Arabic speech recognition tasks with particular emphasis on Arabic dialects, especially the Emirati dialect.
Timeline (last 90 days)
The UAE's Technology Innovation Institute (TII) has launched Falcon-ASR, a speech recognition model focused on the Arabic language.
The Falcon-ASR model has a parameter scale of 1.6 billion.
Falcon-ASR achieved an average Word Error Rate (WER) of 20.92% across 6 Arabic test sets.
Falcon-ASR is optimized with a focus on the Emirati dialect.
TII has officially stated that Falcon-ASR's model weights and API are not yet publicly available.
1.6 billion parameters represents a medium scale among current speech models, balancing recognition capability and deployment cost.
All Facts (6)
The Falcon-ASR model has a parameter scale of 1.6 billion.
50%UnverifiedFalcon-ASR achieved an average Word Error Rate (WER) of 20.92% across 6 Arabic test sets.
50%UnverifiedFalcon-ASR is optimized with a focus on the Emirati dialect.
50%UnverifiedTII has officially stated that Falcon-ASR's model weights and API are not yet publicly available.
50%Unverified1.6 billion parameters represents a medium scale among current speech models, balancing recognition capability and deployment cost.
50%UnverifiedThe UAE's Technology Innovation Institute (TII) has launched Falcon-ASR, a speech recognition model focused on the Arabic language.
50%