Algorithms#

Supplied Processing Algorithms#

SOF contains several permissively-licensed and royalty-free audio processing algorithms that can be used alongside proprietary processing components to build pipelines.

Table 3 SOF Supported Audio Processing Modules & Algorithms#

Algorithm

Source

Category

SIMD

Key Capabilities

Status

Volume / Mute

SOF

Basic Routing & Level

ARM, HiFi 3, HiFi 4, HiFi 5, RISCV, Scalar C

Per-channel linear and log gain curves; Smooth zipper noise attenuation; Zero-overhead bypass when set to 0dB

Upstream

Audio Mixer

SOF

Basic Routing & Level

ARM, HiFi 3, HiFi 5, RISCV, Scalar C

Concurrent playback mixing; Dynamic input stream attachment/detachment; Saturation and clipping protection

Upstream

Sample Rate Converter (SRC)

SOF

Foundational DSP

HiFi 2 EP, HiFi 3, HiFi 4, HiFi 5, RISCV, Scalar C

High SNR polyphase filtering; Low group delay; Multi-channel synchronous resampling

Upstream

Asynchronous SRC (ASRC)

SOF

Foundational DSP

HiFi 3, HiFi 5, RISCV, Scalar C

Farrow polynomial interpolation; Continuous clock drift tracking; Decoupled clock domain bridging

Upstream

Audio Demux

SOF

Basic Routing & Level

Scalar C

Multi-channel stream demultiplexing; Dynamic route splitting; Zero-copy sample extraction

Upstream

Audio Mux

SOF

Basic Routing & Level

Scalar C

Multi-source stream multiplexing; Configurable input channel mapping; Synchronized buffer alignment

Upstream

Channel Map / Remap

SOF

Basic Routing & Level

Scalar C

Arbitrary slot and channel routing; Mono to stereo/surround replication; Channel swap and mute masking

Upstream

Tone Generator

SOF

Diagnostics & Testing

RISCV, Scalar C

Sine wave generation; Configurable frequency and amplitude; Per-channel tone routing

Upstream

Parametric Equalizer (EQ FIR)

SOF

Audio Enhancement

ARM, HiFi 2 EP, HiFi 3, HiFi 4, HiFi 5, RISCV, Scalar C

High-order linear-phase FIR filtering; Speaker and room impulse response correction; Live runtime coefficient updates over IPC

Upstream

Parametric Equalizer (EQ IIR)

SOF

Audio Enhancement

ARM, HiFi 2 EP, HiFi 3, HiFi 4, HiFi 5, RISCV, Scalar C

Cascaded second-order biquad sections; Parametric peak, notch, low/high shelf; Low computational latency

Upstream

Dynamic Range Compressor (DRC)

SOF

Audio Enhancement

ARM, HiFi 3, HiFi 4, RISCV, Scalar C

Speaker excursion and thermal protection; Configurable attack, release, and knee; Peak and RMS signal level detection

Upstream

Multiband DRC

SOF

Audio Enhancement

HiFi 3, HiFi 4, RISCV, Scalar C

Subband crossover splitting; Per-band threshold and ratio controls; Comprehensive speaker protection

Upstream

Crossover Filter

SOF

Audio Enhancement

HiFi 3, RISCV, Scalar C

Linkwitz-Riley 4th order (LR4) splitting; Flat magnitude sum across crossover point; Multi-way woofer, tweeter, and sub routing

Upstream

DC Blocker

SOF

Audio Enhancement

ARM, HiFi 3, HiFi 4, RISCV, Scalar C

Removes DC bias from digital mics and ADCs; Sub-audible rumble attenuation; Near-zero phase distortion in audio band

Upstream

Smart Amp Protection

SOF

Speaker Protection

HiFi 3, HiFi 4, Scalar C

Real-time voice coil temperature estimation; Membrane excursion limiting; Maximizes acoustic output without damage

Upstream

Beamformer (TDFB)

SOF

Voice & Telephony

HiFi 2 EP, HiFi 3, HiFi 4, HiFi 5, RISCV, Scalar C

Multi-mic circular and linear array support; Broadside and endfire steering; Spatial diffuse noise suppression

Upstream

WebRTC Echo Cancellation (AEC)

WebRTC

Voice & Telephony

HiFi 3, HiFi 4, VFPU, Scalar C

Subband adaptive filter convergence; Multi-channel reference loopback alignment; Robust double-talk detection

Active Development

WebRTC Mobile AEC (AECM)

WebRTC

Voice & Telephony

HiFi 3, HiFi 4, VFPU, Scalar C

Fixed-point low-complexity processing; Optimized for earbuds and wearables; Low RAM and cycle footprint

Active Development

WebRTC Noise Suppression (NS)

WebRTC

Voice & Telephony

HiFi 3, HiFi 4, VFPU, Scalar C

Stationary background noise reduction; Configurable aggressiveness levels; Preserves speech formant clarity

Active Development

WebRTC Neural NS (NS2 / RNNoise)

WebRTC

Voice & Telephony

HiFi 3, HiFi 4, VFPU, Scalar C

Recurrent neural network (RNN) inference; Non-stationary transient noise elimination; High speech perceptual quality

Active Development

WebRTC Voice Activity Detector (VAD)

WebRTC

Voice & Telephony

HiFi 3, Scalar C

Multi-band energy likelihood estimation; Sub-frame voice decision gating; Ultra-low power listening states

Active Development

WebRTC Automatic Gain Control (AGC)

WebRTC

Voice & Telephony

HiFi 3, HiFi 4, Scalar C

Dynamic gain adjustment; Saturation prevention limiter; Normalizes quiet and loud speakers

Active Development

Key Phrase Buffer (KPB / WoV)

SOF

Voice & Telephony

Scalar C

Ultra-low power DSP listening mode (D0ix); Zero-latency audio pre-roll buffer playback; Multi-slot capture streaming to host

Upstream

microWakeWord (TFLite Micro)

Google

Voice & Telephony

HiFi 3, HiFi 4, Scalar C

On-device neural network keyword spotting; TFLite Micro runtime execution; Low false-reject and false-alarm rates

Active Development

AAC Decoder

FFmpeg

Codecs & Compression

VFPU, HiFi 3, HiFi 4, HiFi 5, Scalar C

Vector floating-point hardware acceleration; MPEG-4 AAC-LC and HE-AAC profile support; Direct pipeline integration

Active Development

AAC Encoder

FFmpeg

Codecs & Compression

VFPU, HiFi 3, HiFi 4, HiFi 5, Scalar C

Low-power bitstream encoding; Configurable bitrates and sample rates; Optimized MDCT and psychoacoustic model

Active Development

MP3 Decoder

FFmpeg

Codecs & Compression

VFPU, HiFi 3, HiFi 4, HiFi 5, Scalar C

Hardware VFPU SIMD acceleration; High-throughput low-overhead DSP execution; Full bit reservoir and Huffman decoding

Active Development

MP3 Encoder

FFmpeg

Codecs & Compression

VFPU, HiFi 3, HiFi 4, Scalar C

Low-complexity fixed-point encoding; Efficient subband analysis filterbank; Standard MPEG-1 Layer III bitstream generation

Active Development

FLAC Decoder

FFmpeg

Codecs & Compression

VFPU, HiFi 3, HiFi 4, Scalar C

Lossless 16/24-bit audio decompression; Fast linear prediction decoding; Zero fidelity loss playback

Active Development

Opus Decoder

FFmpeg

Codecs & Compression

VFPU, HiFi 3, HiFi 4, Scalar C

SILK speech and CELT music mode support; Sub-20ms algorithmic latency; Dynamic bitrate and bandwidth adaptation

Active Development

Vorbis Decoder

FFmpeg

Codecs & Compression

VFPU, HiFi 3, HiFi 4, Scalar C

General-purpose variable bitrate decompression; Vector quantization floor decoding; Low memory footprint

Active Development

Steam Audio Spatializer

Steam Audio

Spatial Audio

VFPU, HiFi 4, HiFi 5, Scalar C

Spherical 3D sound positioning; Convolution-based HRTF binaural rendering; Dynamic listener and source orientation

Active Development

Real-Time Probes & Telemetry

SOF

Diagnostics & Tools

Scalar C

Direct probe DMA streaming over TCP port 9999; Zero overhead when probe taps are inactive; Multi-point simultaneous stream tapping

Upstream

Note

For detailed algorithm implementation guides, filter tuning workflows, and design tools, consult the Algorithm Specific Information section in Developer Guides.