In　order　to　describe　the　excitation　signal　more　precisely,　an　embedded　speech　coding　algorithm　based　on　ACELP　model　was　proposed.　By　adding　the　number　of　the　pulses　in　algebraic　codebook　and　using　an　innovative　adaptive　codebook　structure,　the　proposed　coding　algorithm　was　favorable　for　finding　the　best　coding　parameters　for　each　layer　and　make　different　layer　independent　relatively.　Compared　to　other　embedded　coding　algorithms　based　on　ACELP　model,　proposed　speech　coder　could　acquire　an　embedded　bit　stream　and　ensure　the　quality　of　the　synthesis　speech　both　on　core　layer　and　the　enhancement　layers.