Digital audio distribution systems have historically operated under severe bandwidth limitations. For decades, listening to radio online meant accepting lossy compression formats like MP3 or Advanced Audio Coding (AAC) at bitrates between 64 kbps and 192 kbps. While these bitrates are sufficient for low-cost playback hardware and casual environments, they discard significant portions of the original audio signal. Lossy compression relies on psychoacoustic modeling to identify and remove frequencies that the human ear is less likely to perceive, particularly quiet sounds that occur immediately after louder signals. This reduction results in a loss of high-frequency detail, a compressed stereo image, and audible coding artifacts during complex musical passages.
High-fidelity broadcasts address these limitations by transmitting audio using the Free Lossless Audio Codec (FLAC). Developed in 2001 by Josh Coalson, FLAC is an open-source, royalty-free digital audio format that provides mathematical reconstruction of the original signal. Unlike lossy formats, FLAC does not discard any audio data. The broadcast stream typically operates at CD-quality resolution, which is 16-bit depth at a 44.1 kHz sampling rate, demanding bitrates between 700 kbps and 1100 kbps. Because the compression process is completely reversible, the decoded analog signal at the listener's end is identical to the original studio recording, preserving the complete dynamic range and frequency spectrum.
The encoding mechanics of FLAC rely on linear prediction and entropy coding. The encoder divides the continuous audio stream into discrete blocks and applies linear predictive coding (LPC) to model the audio waveform. This process estimates each upcoming audio sample based on a linear combination of previous samples. The difference between the actual audio sample and the predicted value is called the residual signal. Because the prediction is highly accurate for musical signals, the residual has a much smaller dynamic range than the raw audio. The encoder then compresses this residual using Golomb-Rice coding, which is an efficient form of entropy coding that assigns shorter bit-sequences to smaller values. This mathematical reduction allows the stream to achieve a compression ratio of approximately 50 percent without any loss of information, applying information theory principles established by Shannon (1948).
To reproduce the full resolution of a FLAC broadcast, the playback chain must support lossless transfer. Standard consumer setups often introduce unexpected compression. For example, common Bluetooth connections rely on lossy codecs such as Subband Coding (SBC) or AAC to transmit audio to headphones. Even if the source stream is lossless, the Bluetooth link compresses the audio, which re-introduces lossy compression artifacts. Listeners must use wired connections to an external Digital-to-Analog Converter (DAC) or transmit the stream over high-bandwidth wireless protocols like Wi-Fi using Universal Plug and Play (UPnP) or AirPlay to maintain the lossless signal to the playback transducer.
Network stability is another important factor when streaming lossless audio. A standard 128 kbps MP3 stream requires minimal network resources, making it resilient to packet loss and high latency. In contrast, a FLAC stream requires over ten times the bandwidth, averaging 1.4 Mbps for uncompressed CD equivalents. If the internet connection suffers from packet loss or jitter, the player buffer can empty, causing audio dropouts. Modern streaming players address this by using larger buffer sizes and adaptive streaming protocols that maintain a continuous audio stream without compromising the playback quality.
References:
Coalson, J. (2001). Free Lossless Audio Codec (FLAC). SourceForge.
Painter, T., & Spanias, A. (2000). Perceptual coding of digital audio. Proceedings of the IEEE, 88(4), 451-513.
Shannon, C. E. (1948). A mathematical theory of communication. Bell System Technical Journal, 27(3), 379-423.