ITU-T Speech Codecs - Specs & Comparison Guide (G.711 to G.729AB)

ITU-T Speech Codecs are voice compression standards defined by the International Telecommunication Union for telephony and IP-based voice services. Consilient Technologies supports a comprehensive range including G.711 (64 kbps, PCM — the universal PSTN codec), G.722 (wideband at 48/56/64 kbps), G.722.1 (wideband at 24/32 kbps), G.723.1 (low bitrate at 5.3/6.3 kbps), G.726 (ADPCM at 16–40 kbps), G.728 (low-delay at 16 kbps), and G.729AB (8 kbps with silence suppression). Each serves a distinct use case in VoIP and telecom infrastructure.

The choice depends on your bandwidth constraints, latency requirements, and interoperability needs. G.711 is the safest default for maximum compatibility across PSTN and SIP networks, but uses 64 kbps per channel. G.729AB is the standard choice for bandwidth-constrained VoIP (SIP trunks, remote offices) at just 8 kbps. G.722 is preferred when wideband/HD voice quality matters and bandwidth is available. G.723.1 targets very low bitrate scenarios like early VoIP gateways. Consilient can help you select and optimize the right codec mix for your specific network conditions.

ITU-T reference code is designed for correctness and conformance testing, not for production deployment. Consilient’s implementations are optimized for real-world use: they include fixed-point variants for embedded processors, platform-specific optimizations for ARM and DSP architectures, reduced memory footprints, and clean APIs designed for integration into SIP stacks, media servers, and embedded firmware. All implementations pass ITU-T conformance testing while delivering measurably lower CPU usage than reference code.

Yes. Consilient’s codec libraries are designed for multi-instance, multi-codec operation — essential for media gateways, session border controllers, and conferencing servers. Each codec instance maintains independent state, and the libraries are thread-safe and re-entrant. This allows you to run hundreds or thousands of concurrent codec channels on a single server, handling transcoding between different codec types (e.g., G.711 to G.729, or G.722 to AMR-WB) in real time.

Several ITU-T codecs support Voice Activity Detection (VAD) and Comfort Noise Generation (CNG) — notably G.729AB and G.723.1. When VAD detects silence, the encoder sends Silence Insertion Descriptor (SID) frames instead of full speech frames, reducing bandwidth by 50–80% during pauses. The decoder generates comfort noise from the SID parameters so the listener still perceives a natural background rather than dead silence. Consilient’s implementations include these features with configurable VAD sensitivity.

Our ITU-T codec portfolio serves telecom carriers (for PSTN/VoIP interworking), VoIP equipment manufacturers (IP phones, ATAs, gateways), conferencing and unified communications vendors, contact center platforms, embedded device makers (intercoms, industrial voice terminals), and lawful interception systems that need to decode captured VoIP streams across multiple codec formats.

Yes. ITU-T codecs are the foundation of SIP-based VoIP — G.711 and G.729 are mandatory in virtually all SIP implementations. Consilient’s libraries integrate directly with SIP media stacks and are deployed in production VoLTE, VoWiFi, and enterprise VoIP environments. We provide sample integration code for common SIP frameworks and RTP media handling.

We provide evaluation packages with pre-built libraries, API documentation, sample code, and ITU-T conformance test vectors for each supported codec. Enterprises can benchmark codec quality using ITU-T P.862 (PESQ) or P.863 (POLQA) measurements, and test CPU/memory performance on their target hardware. Consilient’s engineering team supports integration from evaluation through production deployment.