Opus Audio
Convert Opus audio to MP3, WAV, or OGG directly in your browser — no upload, no server, no data leaves your device. FileDex uses FFmpeg WebAssembly to decode the Opus bitstream from its Ogg container and re-encode to your target format locally.
Your files never leave your device
Common questions
Why is Opus considered the best audio codec?
Opus achieves transparent music quality at 96-128 kbps — roughly half the bitrate MP3 needs for equivalent results. It handles both speech (via SILK mode) and music (via CELT mode) in a single codec, switching seamlessly per-frame. No other codec covers the 6-510 kbps range with comparable quality.
Can I play Opus files on my iPhone?
Yes, since iOS 15. Apple added native Opus decoding in both Safari and the system media framework. Devices running iOS 14 or earlier require a third-party app like VLC. WebRTC-based apps (Discord, WhatsApp) have supported Opus on iOS independently since iOS 11.
What is the difference between .opus and .ogg files?
Both use the Ogg bitstream container. Files ending in .opus contain Opus-encoded audio, while .ogg typically contains Vorbis-encoded audio. The file extension signals which codec is inside, but the container format is identical.
Does FileDex upload my Opus file to a server?
No. FileDex converts Opus files entirely in your browser using FFmpeg compiled to WebAssembly. Audio data never leaves your device — all decoding and encoding runs locally in a sandboxed Web Worker.
Why are Discord voice recordings in Opus format?
Discord uses Opus because it is the mandatory codec for WebRTC and delivers intelligible voice quality at bitrates as low as 16 kbps. This allows Discord to maintain voice quality even on poor network connections while using minimal bandwidth per user.
What makes .OPUS special
Low-latency audio transmission demands a codec that can adapt on the fly, and Opus delivers exactly that. Standardized as RFC 6716 in 2012 by the IETF, Opus is an open, royalty-free codec designed for interactive speech and music over the internet. It operates across bitrates from 6 kbit/s to 510 kbit/s and supports sampling rates from 8 kHz narrowband to 48 kHz full band.
Continue reading — full technical deep dive
Opus seamlessly switches between two internal codecs in real-time: SILK (speech-optimized, derived from Skype's proprietary codec) below 8 kHz and CELT (Constrained Energy Lapped Transform, music-optimized) for full-band content. Mode changes happen every 2.5 ms frame without audible artifacts — a hybrid approach unique among audio codecs. The encoder decides which mode to engage based on signal characteristics, and the decoder handles transitions transparently.
Frame structure and latency
Opus supports frame sizes of 2.5, 5, 10, 20, 40, or 60 ms. The default 20 ms frame balances latency and compression efficiency. At 20 ms frames, algorithmic latency sits at 26.5 ms — roughly half what AAC-LC introduces. For real-time communication, 10 ms frames push total glass-to-glass latency under 50 ms on local networks.
The codec uses a range coder for entropy coding rather than Huffman tables. This gives finer granularity in bit allocation across frequency bands. Each frame is self-contained — the decoder can start from any frame boundary without needing prior state, which makes Opus resilient to dropped packets in RTP streams.
Bitrate performance
| Use case | Recommended bitrate | Quality |
|---|---|---|
| VoIP speech | 16–24 kbit/s | Transparent speech |
| Wideband speech | 24–32 kbit/s | Exceeds AMR-WB |
| Music mono | 64 kbit/s | Near-transparent |
| Music stereo | 96–128 kbit/s | Transparent |
| Archival stereo | 160–256 kbit/s | Indistinguishable from source |
At 64 kbit/s mono, Opus matches or beats AAC-LC at the same bitrate in MUSHRA listening tests conducted by the codec's working group. At 96 kbit/s stereo, most listeners cannot distinguish Opus from the uncompressed original. Opus also supports constant bitrate (CBR), variable bitrate (VBR), and constrained VBR modes — VBR is the default and recommended setting for nearly all use cases.
Container and streaming
Opus frames are carried in Ogg containers for files (.opus extension) or in WebM/Matroska for video muxing. WebRTC mandates Opus support in all compliant browsers — Chrome, Firefox, Safari, and Edge all decode it natively. Discord, Zoom, and Google Meet use Opus as their primary voice codec. Spotify uses Opus for its web player streams at 128 kbit/s (free tier) and 256 kbit/s (premium).
The codec includes built-in forward error correction (FEC). When packet loss is detected, the encoder embeds a low-bitrate copy of the previous frame inside the current packet. At 20% packet loss, FEC recovery keeps speech intelligible without retransmission. Discontinuous Transmission (DTX) is another built-in feature — during silence, the encoder sends tiny comfort-noise frames at roughly 1 kbit/s, saving substantial bandwidth in voice calls where only one person speaks at a time.
When to choose Opus
Use Opus for any streaming, VoIP, or web-based audio workflow. It outperforms Vorbis, MP3, and AAC at equivalent bitrates, especially below 96 kbit/s. The main limitation is ecosystem support in legacy hardware: older car stereos, standalone MP3 players, and some smart speakers still lack Opus decoding. For those targets, AAC or MP3 remains the safer choice. Apple devices gained native Opus playback only in iOS 17 and macOS Sonoma.
FileDex converts to and from Opus entirely in the browser — no upload required.
.OPUS compared to alternatives
| Formats | Criteria | Winner |
|---|---|---|
| .OPUS vs .MP3 | Audio quality at 64 kbps Opus achieves listenable music quality at 64 kbps where MP3 produces severe artifacts. Opus's CELT mode uses wider transform windows and superior psychoacoustic modeling at low bitrates. | OPUS wins |
| .OPUS vs .AAC | Compression efficiency at 96 kbps Opus at 96 kbps is widely considered transparent for music — indistinguishable from the original in listening tests. AAC requires approximately 128-160 kbps to achieve equivalent perceived quality. | OPUS wins |
| .OPUS vs .OGG VORBIS | Voice content at 32 kbps Opus's SILK mode is purpose-built for speech at ultra-low bitrates. Vorbis was designed for music and degrades rapidly below 64 kbps on voice content. | OPUS wins |
| .OPUS vs .MP3 | Hardware device support MP3 plays on every audio device manufactured since 1998. Opus support is limited to software players, WebRTC-capable browsers, and modern streaming platforms. | MP3 wins |
Convert .OPUS to...
Technical reference
- MIME Type
audio/opus- Magic Bytes
4F 67 67 53OggS container. Opus header identified within Ogg stream.- Developer
- Xiph.Org / IETF
- Year Introduced
- 2012
- Open Standard
- Yes — View specification
OggS container. Opus header identified within Ogg stream.
Binary Structure
Opus audio is carried within an Ogg container. Each Ogg page starts with the 'OggS' capture pattern (4F 67 67 53) followed by a 27-byte page header. The first audio page contains the OpusHead identification header: the magic string 'OpusHead' (4F 70 75 73 48 65 61 64), a version byte (0x01), channel count, pre-skip sample count (16-bit LE), input sample rate (32-bit LE — informational only, Opus always decodes at 48 kHz internally), output gain (16-bit LE), and channel mapping family. The second page contains the OpusTags comment header with the magic string 'OpusTags' followed by a vendor string and Vorbis-style key=value metadata pairs. Subsequent pages carry Opus audio packets. Each Opus frame encodes 2.5 to 60 ms of audio. The codec operates in three modes: SILK (speech, 8-16 kHz bandwidth), CELT (music, full bandwidth), or hybrid (SILK + CELT for transition content). Mode selection happens per-frame, allowing seamless switching between voice and music within a single stream.
| Offset | Length | Field | Example | Description |
|---|---|---|---|---|
0x00 | 4 bytes | Ogg Capture Pattern | 4F 67 67 53 (OggS) | Sync pattern identifying the start of an Ogg page. All Opus-in-Ogg files begin with this. |
0x1C | 8 bytes | OpusHead Magic | 4F 70 75 73 48 65 61 64 (OpusHead) | Magic string in the first Ogg page payload identifying this stream as Opus audio. |
0x24 | 1 byte | Version | 01 | OpusHead version. Must be 0x01 for current Opus specification. Decoders reject other values. |
0x25 | 1 byte | Channel Count | 02 | Number of output channels. 1 = mono, 2 = stereo. Opus supports up to 255 channels. |
0x26 | 2 bytes | Pre-skip | 38 01 | Number of samples (at 48 kHz) to discard from decoder output at the start. Accounts for encoder delay. Little-endian. |
0x28 | 4 bytes | Input Sample Rate | 80 BB 00 00 (48000) | Original input sample rate in Hz (little-endian). Informational only — Opus always operates at 48 kHz internally. |
0x2C | 2 bytes | Output Gain | 00 00 | Gain to apply to decoder output in Q7.8 fixed-point dB. Used for ReplayGain-style normalization. |
Attack Vectors
- Crafted OpusHead header with invalid channel mapping family can trigger undefined behavior in decoders that do not validate mapping table bounds
- Malformed Ogg page segment tables can cause buffer over-reads in parsers that trust declared lacing values without size validation
- Opus frames with extreme bandwidth or mode flags can trigger edge cases in the SILK/CELT decoder state machine
Mitigation: FileDex processes Opus files entirely in-browser using FFmpeg WebAssembly sandboxed in a Web Worker. No file data leaves the device. FFmpeg's libopus integration handles malformed frames gracefully with error concealment.