Most free sample packs ship as WAV. Commercial libraries sometimes include AIFF. Archivists tend to reach for FLAC. For a computer-based workflow, these differences rarely cause problems. For hardware samplers, they matter, because most devices have firm opinions about which containers they’ll open.
What’s inside a WAV file
WAV (Waveform Audio File Format) is a Microsoft and IBM spec from 1991, built on the RIFF (Resource Interchange File Format) container. A WAV file is divided into named chunks, each with a four-character identifier, a size field, and a payload.
The two chunks that matter for audio are fmt (sample rate, bit depth, channel count, encoding type) and data (the raw PCM samples). Optional chunks carry additional information: cue for cue points and splice markers, LIST for text metadata like title and artist, and smpl for loop regions.
WAV supports 8-bit through 32-bit integer and 32-bit float audio, mono to multi-channel, at any sample rate. That breadth is why “accepts WAV” doesn’t tell you much by itself. A 22,050 Hz, 8-bit, mono file from a vintage sample CD is a valid WAV file. A 96 kHz, 32-bit float, stereo studio session is also a valid WAV file. Hardware firmware handles only the subtypes it was explicitly written for.
One chunk deserves a specific note: cue . It’s how position markers get embedded in WAV files, and several hardware samplers read them for navigation. The Morphagene uses cue points as splice boundaries. Each cue entry marks where one splice ends and the next begins. If you’re preparing samples with embedded markers, the cue chunk is where that information lives, and not every audio editor writes it in the standard form the hardware expects.
What’s inside an AIFF file
AIFF (Audio Interchange File Format) is Apple’s format from 1988, derived from the IFF (Interchange File Format) structure developed for Amiga systems. Conceptually it mirrors WAV: a chunked container with a format descriptor and raw PCM audio. The main chunks are COMM (sample rate, bit depth, channel count) and SSND (the audio samples themselves).
The structural difference that matters for hardware is byte order. WAV is little-endian, because it was designed for x86 PC hardware. AIFF is big-endian, because it was designed for Motorola 68000-based Macs. Byte order doesn’t affect audio quality at all, but it means reading AIFF requires a different parsing path than reading WAV. Firmware that only implements one path won’t load the other format.
AIFF has its own chunk types for markers and loop points. These are semantically similar to WAV cue points but stored in an incompatible format. If your workflow depends on embedded cue points for splice markers or loop regions, staying in WAV throughout the prep process avoids the conversion risk. AIFF markers don’t translate cleanly to WAV cue points and vice versa.
There is also AIFF-C (AIFC), an extension that supports compressed audio codecs including IMA ADPCM. In practice, when hardware claims AIFF support it means uncompressed AIFF, not AIFC.
Audio quality between WAV and AIFF is identical at the same sample rate, bit depth, and channel count. The container format does not affect the samples themselves.
What’s inside a FLAC file
FLAC (Free Lossless Audio Codec) stores PCM audio in compressed form. The compression is lossless: decoding reconstructs the original PCM data exactly, with no loss in quality. The codec uses linear prediction to model the audio signal and Rice coding to compress the prediction residuals, typically reducing file size by 40–60% compared to uncompressed WAV for typical audio material.
The decode step is the practical issue for hardware. Streaming WAV off an SD card is close to a direct memory read. Playing back FLAC requires a decompression pass before the samples reach the DAC, which takes CPU cycles and introduces overhead in the decode pipeline. For a macOS or desktop application, this cost is negligible. For a hardware sampler running an embedded microcontroller at 100–200 MHz with limited RAM, it’s a meaningful burden.
Most hardware sampler manufacturers have not implemented FLAC support. They ship WAV parsers because WAV is nearly universal in the hardware world, requires no decode step, and streams cleanly off microSD and USB drives. A handful of modules and standalone samplers do support FLAC, but they are the exception: mostly devices running full embedded operating systems rather than bare-metal firmware.
FLAC’s real strength for hardware producers is archiving. A sample library in FLAC takes roughly half the storage of the same library in WAV, with zero quality loss. If you keep your master archive in FLAC and convert to WAV before hardware transfer, you get compact long-term storage and lossless files on your card.
Compatibility across hardware
Where the major devices in SampleStack’s supported library land on format support:
| Device | WAV | AIFF | FLAC |
|---|---|---|---|
| Elektron Digitakt II | yes | yes (via Transfer app) | no |
| Elektron Octatrack MKII | yes | no | no |
| Elektron Model:Samples | yes | yes (via Transfer app) | no |
| 1010music Bitbox mk2 | yes | no | no |
| 1010music Blackbox | yes | no | no |
| Make Noise Morphagene | yes | no | no |
| ALM Busy Circuits Squid Salmple | yes | no | no |
| Roland SP-404MKII | yes | yes | no |
| Akai MPC One / One+ | yes | no | no |
| Qu-Bit Nebulae v2 | yes | yes | yes |
| Tasty Chips GR-1 | yes | yes | yes |
The Elektron devices showing AIFF support handle it at the Transfer app level. The Transfer app accepts WAV, AIFF, and MP3 on input and converts everything to 16-bit, 48 kHz mono before writing to the device’s internal +Drive. The AIFF tolerance lives in the import tool, not the playback firmware. The Octatrack MKII reads files directly off CompactFlash with no conversion step in between, which is why it is WAV-only: there’s nothing to absorb the format mismatch.
The two devices with FLAC support, the Qu-Bit Nebulae v2 and the Tasty Chips GR-1, both run full embedded operating systems. The Nebulae v2 runs on Linux. That OS layer provides the infrastructure FLAC decoding requires. Bare-metal sampler firmware generally doesn’t have the CPU budget or memory model to support it.
File size and metadata trade-offs
WAV and AIFF are the same size for equivalent audio content. A stereo, 48 kHz, 32-bit float file is 384 kB per second in either container. FLAC brings that down to roughly 200 kB per second for typical audio material, though highly dynamic or noise-heavy recordings compress less than tonal material.
For a library of tens of thousands of samples, the difference between WAV and FLAC storage is significant. If you’re maintaining a master archive, FLAC’s compression is worth considering. If you’re prepping files for hardware and want to avoid a conversion step, keeping WAV masters is simpler.
Metadata is trickier. WAV metadata has a fragmented history. The LIST-INFO chunk supports a limited set of predefined fields. ID3 tags are a common addition but are technically non-standard in WAV, and support varies considerably across software. AIFF has its own text chunk with similar limitations. FLAC uses Vorbis comment tags, which are more flexible and consistent within FLAC-native tools, but they don’t map cleanly to WAV when converting. If you tag your FLAC archive with BPM, key, instrument type, or other custom fields and then convert to WAV for hardware, check that your conversion tool handles the metadata mapping. Some tools copy Vorbis comments to ID3 fields in the output file; others discard them silently.
SampleStack maintains its own metadata layer (ratings, tags, notes) that survives format conversion, so tags you assign in the app persist regardless of which export container you choose.
Which format to use
WAV is the right default for a hardware-first workflow. AIFF is fine to keep around if your sources are already in AIFF, particularly with Elektron’s Transfer app, which accepts both and converts on import. Going direct to card (the Octatrack, the Morphagene, most Eurorack modules) means converting to WAV first regardless of the source. FLAC is best treated as an archival format: keep masters in FLAC if storage matters, convert to WAV on the way to hardware.
The container is one piece of the answer. The other is the subtype. The Squid Salmple wants 16-bit, 44.1 kHz, mono WAV. The Morphagene wants 32-bit float, 48 kHz, stereo WAV. Both are valid WAV files, but they’re not interchangeable. A Morphagene reel dropped on the Squid Salmple plays back about 8.8% faster than intended (its firmware assumes 44.1 kHz). A Squid Salmple sample on the Morphagene may not load at all, depending on the firmware version.
The compare formats guide has a cross-device reference for sample rate, bit depth, and channel requirements across the full SampleStack device library.
If you’d rather not handle the conversion by hand, SampleStack reads WAV, AIFF, and FLAC, converts to the exact format your target hardware wants, and writes the right folder layout to your card.