AES-2id: Impulse Response File Formats and Metadata
TL;DR
AES-2id recommends storing impulse responses as WAV files at the measurement sample rate (typically 48 kHz) with 24-bit or 32-bit float resolution. Essential metadata includes: sample rate, measurement date, room identification, source and receiver positions, excitation signal type and parameters, equipment used, ambient temperature and humidity, and any post-processing applied. Proper metadata ensures impulse responses remain useful long after measurement. The document also discusses compatibility with common analysis tools (REW, EASE, ODEON) and convolution reverb plug-ins.
WAV File Specifications
The standard interchange format for impulse responses is the WAV file:
| Parameter | Recommended | Minimum |
|---|---|---|
| Sample rate | 48 kHz (match measurement) | 44.1 kHz |
| Bit depth | 32-bit float | 24-bit integer |
| Channels | Mono (omnidirectional) | — |
| Normalization | Peak at -1 dBFS or calibrated level | — |
32-bit float is preferred because it preserves the full dynamic range without clipping risk during post-processing. 24-bit integer provides 144 dB of dynamic range, sufficient for most room impulse responses.
Essential Metadata
AES-2id recommends documenting the following with each impulse response file:
- Measurement context: Room name, address, dimensions, volume
- Positions: Source (x,y,z) and receiver (x,y,z) coordinates relative to a room reference point
- Equipment: Source model, microphone model, interface model, software version
- Excitation: Signal type (sweep/MLS), duration, frequency range
- Environment: Temperature (°C), relative humidity (%), date and time
- Processing: Deconvolution method, windowing, truncation point, number of averages
- Calibration: Whether absolute SPL calibration was applied, reference level
Metadata Storage Options
Metadata can be stored in several ways:
- WAV file INFO chunks or iXML chunks (embedded in the file)
- Sidecar JSON or XML file with the same base filename
- Structured filename encoding key parameters (e.g., "RoomA_S1R3_48k_10s.wav")
- Database record linking to the file
AES-2id recommends embedded metadata when possible to prevent separation from the audio data.
Interoperability
Common tools that consume IR files:
- REW: Imports mono WAV, supports calibrated level
- EASE, ODEON, CATT: Room simulation software that uses measured IRs for validation
- Convolution reverb: Altiverb, Space Designer, IR1 — any mono/stereo WAV
- SonaVyx: 48 kHz / 24-bit WAV import for analysis and parameter extraction
For frequency response data exchange (not time-domain IR), the FRD format (frequency, magnitude, phase as tab-separated text) is the industry standard used by tools like VituixCAD, ARTA, and Smaart.
Multi-Channel IRs
Binaural IRs use 2-channel WAV (left/right ear). Ambisonic IRs use 4-channel (B-format: W, X, Y, Z) or higher-order formats. AES-2id recommends clearly documenting the channel assignment in metadata.
SonaVyx Export
The IR tool exports impulse responses as 48 kHz / 24-bit WAV files with embedded metadata. Frequency response data is available in FRD format for import into REW or VituixCAD. JSON export includes all measurement parameters for archival. See common errors for pitfalls in file handling and SNR requirements for quality thresholds.
Try It Now
Open this measurement tool in your browser — free, no download required.
Last updated: March 19, 2026