Hi all and Jean-Marc, Firstly thanks a lot to you for helps, at last I accomplished to use speex-1.2beta1 in my speech application. I try to understand speex source code and improve speech quality. The problem in my application (like many people in the list says) is noise. So I've added VAD in preprocessor as following: int param_enabled = 1; int param_disabled = 0; int sampling_rate = 8000; SpeexPreprocessState *preprocess; preprocess = speex_preprocess_state_init(frame_size, sampling_rate); /* frame_size = 160 */ speex_preprocess_ctl(preprocess, SPEEX_PREPROCESS_SET_DENOISE, ¶m_enabled); speex_preprocess_ctl(preprocess, SPEEX_PREPROCESS_SET_VAD, ¶m_enabled); speex_preprocess_ctl(preprocess, SPEEX_PREPROCESS_SET_AGC, ¶m_disabled); /* inputAdd is address of captured input voice (short *). is_speech_in_preprocessor = speex_preprocess(preprocess, inputAdd, null); if (is_speech_in_preprocessor != 0) { speex_encode_int(enc_state, inputAdd, bitsAdd); } VAD in preprocessor is very good, there is not noise outside speech. I don't feel problem so much while continuous speech, but when we speak for short intervals, the start of and end of sentences have clatter (maybe clatter does not meet that sound, there is a defect). To overcome this problem, are SPEEX_PREPROCESS_SET_PROB_START, SPEEX_PREPROCESS_SET_PROB_CONTINUE parameters or if (st->nb_preprocess%200==0) in preprocessor.c (line 608) useful? Thanks a lot in advance, Sincerely, Yasemin -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.xiph.org/pipermail/speex-dev/attachments/20061229/b20adaa9/attachment.htm