An　unequal　coefficients　interframe　predictive　split　vector　quantizer　is　proposed.　This　quantizer　is　used　for　quantizing　the　ISF　(Immittance　Spectral　Frequencies)　parameters　of　wideband　speech.　This　quantizer　is　based　on　interframe　correlation　and　split　vector　quantization.　In　this　quantizer,　the　unequal　coefficients　interframe　prediction　is　firstly　applied　on　the　mean-removed　ISF　parameters　vector　to　be　quantized,　and　then　the　ISF　residual　vector　is　quantized　using　split　vector　quantizer.　Experimental　results　show　that　this　quantization　scheme　achieves　transparent　coding　at　46bits/frame,　and　the　average　spectral　distortion　is　lower　than　that　of　AMR-WB　(ITU-T-G.722.2)　speech　coding　standard.