Creative AI Model or AI service
Metric AI releases Armenian transcription models
Metric AI released OhanASR-v0.1 and a streaming variant, offering Armenian speech recognition for recorded audio and incremental live transcription. [1]
- Model access
- Two variants under CC BY-NC 4.0; non-commercial terms apply. [1]
What the model does
The publisher describes adaptation of a multilingual speech model to Armenian recordings, followed by additional training for streaming. Both variants are linked from the release article. [1]
Why it matters
Researchers and developers can assess dedicated Armenian transcription models for applications where a broad multilingual recognizer may need language-specific adaptation. [1]
Key dates
- Metric AI releases Armenian transcription models [1]
Model limitations
Both models use CC BY-NC 4.0, limiting use to non-commercial purposes under that license. The authors report errors with noisy speech, mixed languages and formal versus colloquial forms. Published evaluation scores are not guarantees for a new recording environment. [1]
Sources and verification
Catalog evidence checked How we verify records
- Primary source Introducing OhanASR: Open Armenian Speech Recognition Models
Related profiles
Creative AI
OhanTTS-v0.1
Metric AI
The model generates audio incrementally and can use a short reference recording with its transcript. The authors provide model access and examples for streaming speech applications.
Creative AI
Amazon Polly voice cloning
Amazon Web Services
Cloned voices run on the Polly Generative engine. Launch support covers seven languages, with each voice speaking the language of its source recording. AWS describes encrypted, account-controlled access to the resulting voice.
Creative AI
Tinker Tailor
Tinker Tailor Creator Inc.
Its Tara assistant helps explore silhouettes, fabrics, colors and prints. Users refine a design in the studio and preview it in 3D before ordering. Production and fulfillment use a library of garment patterns and manufacturing partners.