Transcribe video URLs or audio files to plaintext offline via CPU.
Snap any video URL or audio file into plaintext instantly on your local CPU without cloud APIs.
Yapsnap is a single-command command-line tool that turns local audio files or video URLs from YouTube, TikTok, and X into local plaintext transcripts. It runs entirely on your local CPU using ONNX models without needing a GPU, cloud service, or PyTorch.
It fetches media using yt-dlp, decodes audio with ffmpeg, and processes it locally using streaming Zipformer transducer models. It supports automatic language detection across ten languages, sentence-level timestamps, and optional speaker diarization.
It suits solo practitioners, researchers, and clinicians who need to transcribe recorded client sessions, interviews, or media clips locally to preserve confidentiality. It does not suit users who require a graphical user interface or non-technical staff unfamiliar with the command line.
Transcribes audio quickly on standard CPU cores without requiring CUDA or Apple Silicon.
Accepts links from YouTube, TikTok, X, Instagram, and direct media files via yt-dlp.
Identifies one of ten supported languages using a lightweight whisper-tiny head.
Tags sentences with speaker labels to map who spoke when during multi-person recordings.
Free under its open source license, with no paid hosted option or subscription fees. The only cost is your own local CPU compute and storage for the models.
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
These reviewed AI tools do the same job instantly — from our scored directory.
Only for the first run to download the language identification head and language model, and when fetching media from URLs. Subsequent runs on local files work entirely offline.
No, yapsnap is strictly a command-line tool invoked via your terminal.
It auto-detects from ten supported languages by default, or you can force a specific language code using the --lang flag.
See Zekai first in Google