Free online USM to OPUS converter

What the USM format is

USM is a container used by CRI Middleware's movie systems, especially in video game cutscenes. Games from publishers using CRI tools may store prerecorded movies, Japanese or localized voice tracks, and subtitle-related streams in files such as movie.usm. A USM file can contain video plus ADX or HCA audio, and its internal streams are often interleaved or encrypted for game playback.

General-purpose players and standard audio converters may not recognize USM directly. Tools such as VGMToolbox, USM extraction utilities, and vgmstream are more suitable for inspecting and decoding its embedded streams.

What the OPUS format is

OPUS is a lossy audio codec standardized by the IETF and stored commonly in an Ogg-based .opus file. It supports speech and music at low latency, with configurable bit rates and sample rates, and is supported by browsers, VLC, FFmpeg, Discord, and many current media frameworks.

OPUS is useful when the required result is a compact audio file for playback, web delivery, voice applications, or archival copies where some lossy compression is acceptable. It cannot contain the original USM video stream, subtitles, or game-specific metadata in a generally usable form.

How to actually convert USM to OPUS

1. Make a working copy of the USM file and confirm that you have permission to extract its contents.
2. Extract or demultiplex the embedded audio with a USM-capable tool. In VGMToolbox, use the USM extraction or demultiplexing function to produce the internal audio stream, commonly an .adx or .hca file. Some USM files contain more than one audio stream; extract the language or channel you need.
3. Decode the extracted CRI audio with vgmstream. For example:
   vgmstream-cli -o extracted.wav input.adx
   Replace input.adx with the extracted ADX or HCA filename. If the tool reports unsupported encryption or a missing key, that particular game asset cannot be decoded without the required game-specific support.
4. Encode the decoded WAV file as OPUS with FFmpeg:
   ffmpeg -i extracted.wav -c:a libopus -b:a 128k output.opus
   Use approximately 64k for speech, 96k to 128k for general stereo audio, or 160k to 192k when preserving more music detail matters. FFmpeg can also read a supported extracted format directly, for example:
   ffmpeg -i input.adx -c:a libopus -b:a 128k output.opus
5. Play output.opus and verify its duration, channel layout, language, and synchronization. If the USM contained separate audio streams, repeat the process for each extracted stream.

You can also try converting the file for free on 101convert.com without installing software, provided the service accepts the specific USM variant and its embedded audio stream.

Quality and compatibility limitations

The conversion may be lossy twice: CRI audio can already be compressed as ADX or HCA, and OPUS applies another lossy encode. Encoding from the decoded WAV avoids an unnecessary intermediate compression step, but it cannot restore detail lost in the original game asset.

Choose -ac 1 only when a mono result is required; otherwise FFmpeg preserves the source channel count when the encoder supports it. Use -ar 48000 only when a fixed 48 kHz output is needed, since OPUS internally targets 48 kHz and FFmpeg can resample automatically.

If the objective is to retain the cutscene, convert the video stream to a video container such as MP4 or Matroska instead; OPUS alone is not a video replacement.