cricodecs 1.1.2

CriWare codec and container library with Python bindings.

CriCodecs and CriStudio

CriStudio is the desktop interface for inspecting, previewing, extracting, editing, and building CRI middleware files. It is powered by CriCodecs, a C++23 codec and container library with a standalone CLI and Python bindings.

Most users should download CriStudio. The CLI, Python package, and C++ SDK are available for automation, scripting, and application integration.

Inspect and play muxed VP9 and ADX streams directly from a USM container.

Browse nested CPK entries and inspect structured binary data without extracting it first.

Download

The latest GitHub release provides platform-specific downloads:

Download	Intended use
CriStudio portable	Recommended. Extract and run; Qt and FFmpeg are included.
CriStudio slim	Smaller advanced-user build; requires compatible Qt and FFmpeg on the system.
CriCodecs CLI	Optimized command-line executable for scripts and batch work.
C++ SDK	Headers, CMake package metadata, and static/shared libraries.
Python wheels	Install from PyPI with python -m pip install cricodecs.

Portable CriStudio needs no installer. Windows packages expose only CriStudio.exe and an _internal runtime directory, Linux uses an AppImage, and macOS uses a self-contained application bundle.

CriStudio

CriStudio has two workspaces:

• Browse loads files and archives, searches nested entries, previews audio, video, images, tables, and raw bytes, and extracts selected content.
• Editor edits archive entries and tables, builds and muxes media, and manages encryption keys without changing the global CRI key implicitly.

Key recovery is available for supported HCA, ADX, AHX, USM, AWB, and ACB inputs. Multi-file recovery runs in the background and ranks a bounded set of candidates rather than blocking the interface.

Format coverage

Area	Formats
Audio and audio containers	ADX, AHX, HCA, WAV/PCM, AAX, AIX
Sound banks and tables	ACB, AWB, ACX, CSB, UTF
Archives and disc containers	AFS, CPK, CVM/ROFS
Video and stream containers	USM, SFD/SofDec

Editing, encoding, muxing, and key-recovery actions appear only where supported.

CLI

Release builds provide a standalone cricodecs executable. Python installs also provide a cricodecs console command backed by the same native implementation.

Common usage is input-driven:

cricodecs input.hca
cricodecs input.cpk

Audio inputs such as ADX, AHX, HCA, and AAX decode to WAV by default. Archive and container inputs extract to a sibling directory by default. Use -m for metadata, -m --json for JSON metadata, -f to force the input format, and -o to choose the output path or extraction root.

Key recovery

cricodecs --recover-key -f hca music.hca
cricodecs --recover-key -f hca music.acb
cricodecs --recover-key -f usm movie.usm
cricodecs --recover-key -f adx music.adx
cricodecs --recover-key -f ahx voice_a.ahx voice_b.ahx
cricodecs --recover-key -f awb BGM.awb
cricodecs --recover-key -f acb BGM.acb --json

HCA recovery scans standalone files, AWB/ACB/USM containers, and folders. USM recovery combines supported video, encrypted HCA, and masked ADX evidence. ADX/AHX recovery returns directly usable start,mult,add triplets, while AWB/ACB recovery targets encrypted AAC/M4A entries. Multiple inputs are assumed to share one base key unless --independent is supplied.

HCA type 56 uses only bits 0 through 55, so recovery cannot determine the original 64-bit key's upper byte. If the exact stored value matters, the seven recovered bytes can be searched in application metadata or executables, but a match is only a heuristic; the key may instead be derived, obfuscated, split, or absent.

USM muxing

When building a USM, --audio accepts repeatable ADX or HCA inputs. Supplying --key masks the video and ADX audio by default, converts plain HCA audio to cipher type 56, and preserves HCA that is already encrypted:

cricodecs --build -f usm --key 0x165CF4E2138F7BDA \
  --audio dialogue.adx --audio music.hca -o movie.usm movie.ivf

See USAGE.md for the complete command and key-recovery options.

Python Usage

Install from PyPI and import the package as cricodecs:

python -m pip install cricodecs

The top-level cricodecs.load() helper accepts a path or bytes and returns the matching object. Individual modules expose operations such as load(), decode(), encode(), extract(), demux(), and save_bytes().

Top-Level Load

import cricodecs

obj = cricodecs.load("input.cpk")
print(type(obj))
print(repr(obj))

ADX Decode And Encode

from pathlib import Path

from cricodecs import adx

wav_bytes = adx.decode("input.adx")
Path("output.wav").write_bytes(wav_bytes)

encoded = adx.encode(Path("source.wav").read_bytes())
Path("output.adx").write_bytes(encoded)

HCA Decode, Encode, Encrypt

from pathlib import Path

from cricodecs import hca

wav_bytes = hca.decode("input.hca", keycode=0xCF222F1FE0748978)
Path("decoded.wav").write_bytes(wav_bytes)

config = hca.HcaEncodeConfig()
config.sample_rate = 48000
config.channel_count = 2
config.quality = hca.HcaQuality.HIGH

hca_bytes = hca.encode(Path("source.wav").read_bytes(), config)
Path("encoded.hca").write_bytes(hca_bytes)

encrypted = hca.encrypt(hca_bytes, cipher_type=56, keycode=0xCF222F1FE0748978)
Path("encrypted.hca").write_bytes(encrypted)

CPK Inspect, Extract, Build

from pathlib import Path

from cricodecs import cpk

archive = cpk.load("input.cpk")
print(archive.info())
archive.extract("input_extracted")

new_archive = cpk.create(cpk.CpkPreset.FILENAME)
new_archive.add_file("data/voice.adx", "voice/voice.adx")
new_archive.add_bytes(b"hello", "text/readme.txt")
Path("output.cpk").write_bytes(new_archive.save_bytes())

ACB/AWB And USM

from cricodecs import acb, usm

cue_sheet = acb.load("sound.acb")
print(cue_sheet.info())
cue_sheet.extract("sound")

movie = usm.load("movie.usm")
print(movie.info())
movie.extract("movie_streams")

config = usm.UsmMuxConfig(
    video_path="movie.264",
    audio_tracks=[usm.UsmMuxAudioTrack("movie.adx")],
    subtitle_tracks=[
        usm.UsmMuxSubtitleTrack(
            "subtitles_en.srt",
            language_id=0,
            format=usm.UsmSubtitleFormat.SRT,
        ),
    ],
)
usm.mux(config, "movie_with_subtitles.usm")

Objects and configuration structs provide useful repr() output:

from cricodecs import adx

print(repr(adx.AdxEncodeConfig()))

C++ Usage

Include the complete API and link the namespaced CMake target:

#include <cricodecs/cricodecs.hpp>

find_package(CriCodecs CONFIG REQUIRED)
target_link_libraries(my_app PRIVATE CriCodecs::CriCodecs)

Module headers are installed under <cricodecs/...>, so projects can include only the formats they use. Public operations report recoverable failures with std::expected<T, std::string>.

ADX

#include <filesystem>
#include <fstream>
#include <vector>

#include <cricodecs/adx/adx_codec.hpp>

int main() {
    auto adx = cricodecs::adx::Adx::load("input.adx");
    if (!adx) {
        return 1;
    }

    auto decoded = adx->decode();
    if (!decoded) {
        return 1;
    }

    const auto& pcm = decoded->pcm_data;
    (void)pcm;
}

HCA

#include <cstdint>
#include <vector>

#include <cricodecs/hca/hca_codec.hpp>

std::vector<int16_t> decode_hca(std::span<const uint8_t> bytes) {
    auto decoded = cricodecs::hca::decode(bytes, 0xCF222F1FE0748978ULL);
    if (!decoded) {
        return {};
    }
    return std::move(*decoded);
}

CPK

#include <cricodecs/cpk/cpk_container.hpp>

int main() {
    auto archive = cricodecs::cpk::Cpk::load("input.cpk");
    if (!archive) {
        return 1;
    }

    return archive->files().empty() ? 1 : 0;
}

Build from source

Requirements

• CMake 4.2 or newer
• A C++23 compiler:
  • GCC 16.1 or newer on Linux
  • Visual Studio 2026 / MSVC v145 or newer on Windows
  • A current Apple Clang/Xcode toolchain on macOS
• Python 3.9 or newer when building the Python package

Use a fresh build directory after changing the compiler, generator, Python version, or CMake version.

Python package

From the repository root:

python -m pip install .

For an editable development install:

python -m pip install --no-build-isolation -ve .

C++ SDK and CLI

Build and install an optimized static library and the CLI:

cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release
cmake --build build --parallel 4
cmake --install build --prefix ./install

Set -DBUILD_SHARED_LIBS=ON for a shared library. To omit the CLI, set -DCRICODECS_BUILD_CLI=OFF.

On Windows with Visual Studio 2026:

cmake -S . -B build-vs -G "Visual Studio 18 2026" -A x64
cmake --build build-vs --config Release --parallel 4
cmake --install build-vs --config Release --prefix install

Credits

CriCodecs builds on public knowledge and prior open-source work around CRI formats. Many thanks and credits to:

• vgmstream for HCA and CRI audio research.
• VGAudio for ADX and HCA codec behavior references.
• bnnm for CRI audio-format research and tooling.
• Nyagamon for CRI format research.