Digital phone audio, which was μ-law [1] rather than linear PCM, was not exactly uncompressed. However, the compression involved was very simple, and was designed to offer the 8-bit samples a higher dynamic range than linear PCM would be capable of at the same bitrate (linear PCM would need 14 bits to represent the full dynamic range of 8-bit μ-law). It resulted, by design, in quite clear audio for human voice, as compared to the more aggressive compression used today, but couldn't handle non-voice audio like music as well.

[1] μ-law was used in North America and Japan, but a similar encoding called A-law was used elsewhere