Just rolled out Call Insights in Tish.
After every call you get what's in the screenshot: Talk/Listen, NC effectiveness in dB, noise floor, speech hygiene (lip smacks, chewing, background events).
Aleksandr Beshkenadze
@beshkenadze
- 54 Followers
- 1 Following
- 25 Posts
Startup Founder | AI Product Architect | Building Privacy-First Legal & EdTech Tools (RAG, Hybrid Search, MCP Servers) | Node.js • TypeScript • Flutter • DevOps
Shipping before summer. Beta is open, 50% off Pro for testers:
→ https://tish.bshk.app/?utm_source=twitter&utm_medium=social&utm_campaign=call_insights_article&utm_content=aleksandr_post
→ https://tish.bshk.app/?utm_source=twitter&utm_medium=social&utm_campaign=call_insights_article&utm_content=aleksandr_post
Runs automatically the moment the call ends. Not happy with the result? One Reanalyze button.
Under the hood: VAD in parallel on raw + processed audio, SED for hygiene, LUFS from preview — single pass over the files.
Every metric answers a concrete question: "was I cutting people off," "did the noise canceller actually do work," "was I crunching chips into the mic." Score 100 isn't "we're awesome" — it means speech hygiene was in range.
Currently polishing things up before an App Store release. If you work with audio and want to give it a try — reach out, I'd love the feedback.
#audio #podcasting #macos #indiedev #audioprocessing #noisereduction #solovey #buildinpublic
Solovey fixes that in a couple of clicks: drop in a file, the app removes noise, levels out the volume — and gives you a clean result. Everything runs locally on your Mac, no cloud uploads.
Speed-wise: a 10-minute recording processes in about 3 minutes on a MacBook Pro.
Building Solovey — a macOS app that cleans up audio recordings
You know the deal: you recorded a podcast, interview, or lecture — and the audio has background noise, hum, and volume jumping all over the place.
**11/**
Want to adapt this for your domain?
The approach works for any specialized translation:
• Legal
• Medical
• Technical
• Gaming
Build a dictionary. Fine-tune a small model. Beat the giants.
**10/**
Key lessons:
1. Remove what you don't need
2. Domain dictionaries > model size
3. 16-bit LoRA >> 4-bit QLoRA
4. Measure everything
5. Iterate relentlessly
**9/**
Final results:
• 400K recipes in 2.5 hours
• RTX 4090
• $5 electricity
• 90% quality
• 155ms per translation
DeepL would need 55 hours and $5,000.