Opus Bitrate Guide: 32, 64, 96, 128, 192 kbps Explained

Opus is the most efficient general-purpose audio codec in widespread use. Picking the right bitrate is the single biggest decision you make when encoding Opus, and the right answer is almost always lower than you would use with MP3 or AAC. This guide gives concrete numbers for each bitrate band, the application-mode flag that affects them, and the encoder behaviors that make Opus deliver more quality per kilobit than its predecessors.

Opus Bitrate Range

Opus operates from 6 kbps up to 510 kbps per channel. In practice the useful range is roughly 16 to 256 kbps. The codec is VBR by default — the bitrate flag sets a target average, and the actual instantaneous rate varies with content complexity. The encoder is also intelligent about silence and low-content passages, frequently using far fewer bits than the target during quiet moments.

Application Modes Affect Quality at the Same Bitrate

Before bitrate, the application mode shapes everything:

'voip' mode prioritizes SILK, Opus's speech-optimized layer. Best for voice content; significantly better intelligibility at 16-32 kbps than 'audio' mode.
'audio' mode (default) balances SILK and CELT, automatically picking based on content. Best for mixed content, music, and general audio.
'lowdelay' mode disables SILK and uses CELT only for minimum algorithmic latency. Best for interactive audio (WebRTC, game voice).

The same 32 kbps target sounds dramatically different in voip vs audio mode for speech content. Pick the mode before fine-tuning the bitrate.

Bitrate Bands

16-24 kbps: voice-only, ultra-low bandwidth

In voip mode, Opus at 16-24 kbps mono produces intelligible speech that survives noisy connections. This is the bitrate range used by WhatsApp voice notes (16 kbps) and Telegram voice messages (~32 kbps). Music at this bitrate is unlistenable; use only for voice content under bandwidth constraints.

32 kbps: voice excellent, music thin

In voip mode at 32 kbps mono, voice is clean and natural. Sibilance and consonant clarity are well-preserved. Use for podcasts, audiobooks, voice-over narration where bandwidth matters. Music at 32 kbps lacks bass and sounds compressed; do not use for music.

48-64 kbps: speech transparent, music acceptable

48 kbps mono in voip mode produces voice essentially indistinguishable from much higher bitrates for typical speech content. 64 kbps stereo in audio mode handles music with audible compression on dense passages but acceptable quality for casual listening. Discord voice channels run around 64 kbps.

96 kbps: the music sweet spot

The practical target for general music delivery on the web. At 96 kbps stereo in audio mode, most listeners cannot identify the audio as compressed in casual listening. Artifacts appear only on critical listening through reference-quality headphones on demanding source material (string quartets, dense orchestral, sibilance in solo vocals). For web audio players, streaming apps, web games, and any music delivery where file size matters, 96 kbps Opus is the recommended default.

128 kbps: transparent for nearly everyone

At 128 kbps stereo, Opus is effectively transparent in ABX testing for the vast majority of listeners and material. This is the bitrate above which improvements are detectable only by trained listeners on specific reference tracks. For serious web delivery, app audio, or anywhere quality cannot be questioned, 128 kbps is the safe upper-quality target.

192 kbps: audiophile delivery

At 192 kbps stereo, Opus exceeds what virtually all listeners can detect as different from lossless source. Use only when storage is irrelevant and you want zero-doubt quality (audiophile delivery, mastering reference, archival lossy backup). Diminishing returns above this — 256 kbps Opus is essentially identical in perceived quality to 192 kbps for nearly all material.

VBR Behavior and Predictability

Opus is VBR by default. Setting '-b:a 96k' targets a 96 kbps average; the actual file's average lands within ~10% on typical music, with quiet passages encoding at 50-70 kbps and loud complex sections reaching 130-150 kbps.

To force CBR (rare; useful for fixed-bandwidth streams):

'ffmpeg -i input.wav -c:a libopus -b:a 96k -vbr off output.opus'

CBR Opus is slightly worse perceptually than VBR Opus at the same average bitrate. Leave VBR on unless the delivery channel demands fixed rate.

ffmpeg Flags

Common encode patterns:

'ffmpeg -i input.wav -c:a libopus -b:a 96k output.opus'

— general music at 96 kbps stereo.

'ffmpeg -i input.wav -c:a libopus -b:a 32k -ac 1 -application voip output.opus'

— voice content at 32 kbps mono with voip optimization.

'ffmpeg -i input.wav -c:a libopus -b:a 96k -application lowdelay output.opus'

— interactive audio with minimum latency.

'ffmpeg -i input.wav -c:a libopus -b:a 192k -compression_level 10 output.opus'

— maximum quality encode at 192 kbps with the highest compression effort (slowest encode, smallest file at quality target).

The 'compression_level' flag (0-10, default 10) trades encode speed for efficiency. Decode speed is unaffected.

Practical Recommendations Summary

Voice-only voicemail or notification audio: 16-24 kbps mono voip.
Podcast voice: 32-48 kbps mono voip.
Web music background: 64-96 kbps stereo audio mode.
General music delivery: 96-128 kbps stereo audio mode.
High-fidelity music: 128-192 kbps stereo audio mode.
Real-time game voice: 32-64 kbps lowdelay mode.
Real-time game music/SFX: 64-128 kbps lowdelay mode.

Comparing to MP3 and AAC

Opus is roughly 30-50% more efficient than MP3 and 20-30% more efficient than AAC at low and medium bitrates. Practical equivalences:

96 kbps Opus ≈ 192 kbps MP3 ≈ 128 kbps AAC.
128 kbps Opus ≈ 256 kbps MP3 ≈ 192 kbps AAC.
192 kbps Opus ≈ 320 kbps MP3 ≈ 256 kbps AAC.

If you are migrating from MP3 settings, halve the bitrate as a starting point and let your ears confirm. For deeper format-level coverage, see Opus vs MP3 and Opus vs AAC. The What is Opus explainer covers the SILK+CELT design that makes the efficiency possible.