Elixir bindings for the Vosk API speech recognition library.

VoskEx provides offline, high-performance speech-to-text capabilities for Elixir applications through bindings to the Vosk API.

Installation

Add vosk_ex to your list of dependencies in mix.exs:

def deps do
  [
    {:vosk_ex, "~> 0.1.0"}
  ]
end

VoskEx automatically downloads precompiled Vosk libraries during compilation, so no system dependencies are required!

Simply run:

mix deps.get
mix compile  # Automatically downloads Vosk library (~2-7 MB) for your platform

Supported platforms:

• Linux: x86_64, aarch64 (ARM64)
• macOS: Intel (x86_64), Apple Silicon (M1/M2/M3)
• Windows: x64

Configuration

VoskEx logs are disabled by default to keep your application logs clean. To enable Vosk/Kaldi internal logging, add to your config/config.exs:

config :vosk_ex,
  log_level: 0  # -1 = silent (default), 0 = default logging, >0 = more verbose

Quick Start

# Download a model first
Mix.Task.run("vosk.download_model", ["en"])

# Load model and create recognizer
{:ok, model} = VoskEx.Model.load("models/vosk-model-small-en-us-0.15")
{:ok, recognizer} = VoskEx.Recognizer.new(model, 16000.0)

# Process audio (PCM 16-bit mono)
audio = File.read!("audio.wav") |> binary_part(44, byte_size(audio) - 44)
case VoskEx.Recognizer.accept_waveform(recognizer, audio) do
  :utterance_ended ->
    {:ok, result} = VoskEx.Recognizer.result(recognizer)
    IO.inspect(result["text"])
  :continue ->
    {:ok, partial} = VoskEx.Recognizer.partial_result(recognizer)
    IO.inspect(partial["partial"])
end

Architecture

This module provides low-level bindings to the Vosk C API. For a more user-friendly interface, use VoskEx.Model and VoskEx.Recognizer.

The library uses a three-layer architecture:

• Layer 1 (C NIF): Direct bindings to Vosk C API via erl_nif.h
• Layer 2 (Low-Level): This module - thin Elixir wrapper with NIF stubs
• Layer 3 (High-Level): VoskEx.Model and VoskEx.Recognizer - idiomatic Elixir API

Resources are automatically managed through BEAM garbage collection.