Skip to content
pubscout

whisper_ggml

2.10.0, released 4 days ago by antonkarpenko.com

On-device speech-to-text for Flutter powered by whisper.cpp

$ flutter pub add whisper_ggml
Source repository

What it does

Provides on-device automatic speech recognition (ASR) using Whisper models via whisper.cpp. Supports both one-off audio file transcription and live streaming transcription with optional VAD and speaker-turn detection.

AI-written summary of version 2.9.0, based on the package's README and metadata. Report an inaccuracy

Use it when

  • You need completely offline speech-to-text without cloud APIs or keys.
  • You want live streaming speech transcription in a Flutter app.
  • You need voice activity detection (VAD) or basic speaker diarization.

Skip it when

  • You require cloud-based speech recognition services with massive pre-trained language vocabularies.
  • You cannot accept the app bundle size increase from including or downloading Whisper model weights.

Strengths

  • Fully offline, local execution with no server dependency.
  • Supports live streaming transcription with partial results.
  • Fast native performance utilizing whisper.cpp with hardware acceleration.

Weaknesses

  • Requires managing large model files (from tens of megabytes to gigabytes).
  • Platform configuration required for native builds and minimum OS targets (e.g. iOS 15.6).

Setup

Kind
Plugin (native code)
Effort
Medium
Code generation
Yes (build_runner)
Direct dependencies
7

Requires setting the iOS deployment target to 15.6 or later in Podfile and Xcode. On Windows and Linux, FFmpeg must be installed on PATH if non-WAV audio files are processed.

import 'package:whisper_ggml/whisper_ggml.dart';

final controller = WhisperController();

final result = await controller.transcribe(
  model: WhisperModel.tiny,
  audioPath: '/path/to/audio.wav',
  lang: 'en',
);

print(result?.transcription.text);

Alternatives

Packages that do the same job and could replace whisper_ggml.

Works with

Packages commonly used together with whisper_ggml: ones it builds on, ones built on it, and its own family.