Codec-SUPERB is a comprehensive benchmarking framework designed to evaluate sound codec models across various applications and signal-level metrics. The framework aims to address the lack of a unified evaluation standard for sound codecs, which have been primarily assessed based on signal-level metrics without considering downstream applications. Codec-SUPERB provides a platform for developers and users to collaborate, share results, and compare codec models through an online leaderboard. It includes diverse datasets spanning speech, audio, and music, and evaluates codecs from both application and signal perspectives. The framework also introduces a unified overall score that integrates multiple signal-level metrics, enabling a more comprehensive assessment of codec performance. The study analyzes various sound codec models, including Speech-Tokenizer, AudioDec, AcademiCodec, Descript-audio-codec, Encodec, and FunCodec, and evaluates their performance across different datasets and applications. The results show that DAC achieves a balanced trade-off between performance and bitrate, while Academicdec excels in maintaining performance at lower bitrates. The study also highlights the importance of considering both signal-level and application-level metrics in evaluating sound codecs. The framework aims to accelerate progress in the codec community by providing open-source resources, including code, leaderboard, and datasets.Codec-SUPERB is a comprehensive benchmarking framework designed to evaluate sound codec models across various applications and signal-level metrics. The framework aims to address the lack of a unified evaluation standard for sound codecs, which have been primarily assessed based on signal-level metrics without considering downstream applications. Codec-SUPERB provides a platform for developers and users to collaborate, share results, and compare codec models through an online leaderboard. It includes diverse datasets spanning speech, audio, and music, and evaluates codecs from both application and signal perspectives. The framework also introduces a unified overall score that integrates multiple signal-level metrics, enabling a more comprehensive assessment of codec performance. The study analyzes various sound codec models, including Speech-Tokenizer, AudioDec, AcademiCodec, Descript-audio-codec, Encodec, and FunCodec, and evaluates their performance across different datasets and applications. The results show that DAC achieves a balanced trade-off between performance and bitrate, while Academicdec excels in maintaining performance at lower bitrates. The study also highlights the importance of considering both signal-level and application-level metrics in evaluating sound codecs. The framework aims to accelerate progress in the codec community by providing open-source resources, including code, leaderboard, and datasets.