Whisper Keeps Correcting Nigerian Speech. Here's How I Measured It
What I Benchmarked I've been fine-tuning Whisper on Nigerian speech for a while, and I built this benchmark to measure my model against the standard…
DevOps
⚑ Сообщить о проблемеПоследние новости DevOps от Tech News
Все темы
agents
ai
api
architecture
automation
aws
backend
beginners
cloud
database
devchallenge
devops
docker
javascript
kubernetes
llm
machinelearning
mcp
node
opensource
performance
productivity
programming
python
security
showdev
testing
tutorial
typescript
webdev
Как я в одиночку написал сервис для расшифровки видео — и три раза чуть не забросил из-за одной и той же ошибки
Дисклеймер сразу: я не собирался писать «ещё один стартап». Мне просто было лень руками пересматривать видео, чтобы выписать из него тезисы. Дальше…
Как нейросеть переводит видео. Часть 1: Whisper врёт, и мы это знаем
Я один пишу сервис дубляжа видео с 2022 года, когда OpenAI выложил Whisper и мне показалось, что дело на месяц. В 2026-м через конвейер прошло 9 203…
How I pick a voice input setup for PC typing (2026)
How I pick a voice input setup for PC typing (2026) I type a lot on a laptop — chat replies, long AI prompts, the occasional email in another langua…
Whisper.cpp Vulkan on Arch: A Detective Story With No Crime
A six-week journey through source builds, CI pipelines, and one package pacman never mentioned. TL;DR: pacman -S whisper-cpp ggml-vulkan . That's it…
Defects Missed in Transcription — AI Speaks After 0.5-Second Silence
📝 Originally published (in Japanese) at forge.workstyle.tech . Quality Control for TTS Models: Why Transcription Isn’t Enough I used to perform qual…
Why AI Models Sometimes Elongate Their Greetings Like 'Kon'nichiwa~'
📝 Originally published (in Japanese) at forge.workstyle.tech . When I had the trained voice model read "こんにちは" (Hello), it stretched the phrase to "…
From Zzz's to Data: Building an AI-Powered Sleep Apnea Monitor with Whisper-v3
Sleep is the ultimate black box. We spend a third of our lives doing it, yet we have almost zero data on what happens during those eight hours—unles…
Practical Techniques to Improve Faster-Whisper Recognition Accuracy — Use Recent Conversations, Not Dictionaries, for initial_prompt
📝 Originally published (in Japanese) at forge.workstyle.tech . Techniques That Actually Improved Accuracy When Running Faster-Whisper as STT for Rea…
DaDaScribe API: YouTube Multi-Language Transcript + SRT in One Request
Most transcription APIs still force you to do this: Download the YouTube video Extract the audio with yt-dlp + ffmpeg Upload the file Call the trans…
Streaming ASR vs Whisper on mobile: when to switch
The problem: Whisper feels laggy for live apps Voice-first apps became mainstream in 2024-2026. Voice agents, real-time captions, live meeting trans…
Snoring Secrets: Fine-Tuning Whisper-v3 to Identify Sleep Apnea Events Like a Pro
Have you ever wondered if that loud snoring is just a nuisance or a genuine health red flag? Sleep Apnea detection is traditionally done in uncomfor…
Я ошибался: бенчмарк 23 ASR-нейросетей для русской айтишной диктовки
В марте я советовал Whisper Large v3 для голосовой диктовки и ошибался дважды: в методе (выбирал модель «на ощупь») и в самой модели. Пересобрал всё…
The Culprit Behind the 'Slow Speech' Bug in Voice Conversion Was Whisper's 30-Second Limit
📝 Originally published (in Japanese) at forge.workstyle.tech . While building a voice conversion app, I received a report like this: "The recording…
Stop Sending Your Snores to the Cloud: Build a Privacy-First Sleep Guardian with Whisper-tiny and TCN on Raspberry Pi
Let’s be honest: no one wants their private nighttime "soundtrack" (a.k.a. snoring or heavy breathing) being uploaded to a corporate server for "ana…
How to Run Voice-to-Text Locally on Your Desktop (Whisper, Offline Dictation)
OpenAI trained Whisper on 680,000 hours of audio, and the small models that came out of it run on a laptop CPU in real time. That power sits idle wh…
How to Transcribe Meetings Locally in 2026 (Whisper, On-Device)
OpenAI's Whisper model runs accurate speech-to-text on a laptop GPU, and the C++ port runs it without Python or a cloud account. That capability sit…
Qwen3.5 на двух V100, reverse SSH вместо Cloudflare в Telegram Mini App: собираю AI-репетитора английского
У меня в углу комнаты стоит сервер с двумя Tesla V100 32GB. Они доcтались мне для другой задачи, которая отвалилась, и полгода стояли мёртвым грузом…
SpeakShift: A Fully Local Desktop App Powered by Whisper.cpp + NLLB + FFmpeg
SpeakShift: Fully Local Whisper.cpp + NLLB Translation + FFmpeg Media Converter Hi DEV Community 👋 Like many of you, I spend a lot of time working w…