Hey everyone!
I built Telvr — a desktop app that lets you dictate text with a single hotkey and inserts it directly at your
cursor, in any app. No copy-paste, no app switching. Available on macOS and Windows.
How it works
- Press your hotkey
- Speak
- Release — your text appears where your cursor is, in under 2 seconds
It works system-wide: emails, Slack, code editors, browsers — anywhere you can type.
How Groq powers Telvr
Telvr relies entirely on Groq for the AI pipeline:
- Whisper Large v3 for speech-to-text (~1.2s latency — this is where Groq’s speed really shines)
- LLaMA 3.3 70B for optional AI enrichment of the transcribed text
The enrichment step transforms raw speech into different output formats — clean text, professional emails, meeting notes, summaries, or
custom formats with your own prompts.
Why Groq?
Latency is everything for voice input. When you release the hotkey, you expect text to appear now — not in 5 seconds. Groq’s inference
speed makes the whole experience feel instant. Whisper on Groq consistently returns in ~1.2s, which is fast enough that the flow feels
natural. No other provider comes close for this use case.
Platforms
- macOS (Apple Silicon + Intel)
- Windows
Website: telvr.ai
Would love to hear your feedback or answer any questions!