Results 1 to 3 of 3
  1.    #1  
    I just learned that the
    GSM Treos use the AMR audio codec and the
    CDMA Treos use the QCELP audio codec.

    Since they are not the same which is better.

    As I understand it, the AMR codec focuses on reproducing the narrow frequncy range of the voice leaving outlying frequncies to falter (which is disappointing for this GSM user). Am I off base hear? How does this differ from those Sprint Treo 650s?

    Any techno geeks...
  2. #2  
    Well, comparing CDMA codecs to AMR isn't as easy as you make it out to be. For one thing, Sprint doesn't use QCELP by default, per se, though it will fallback to QCELP if you have a way older phone that can't use any other vocoder.

    There are three flavors of vocoders in CDMA: 13kbps QCELP, 8kbps QCELP, and EVRC. Most phones sold on Sprint today can do all three modes, but the default setting is EVRC. The differences are:

    1. 8k QCELP uses scalar linear quantization, and samples audio in a single phase. In layman's terms: it's heavily compressed and range is limited.

    2. 13k QCELP uses vector quantization, and a dual-stage sampling process. In layman's terms: If it's done right, 13k QCELP can exceed toll quality. Anyone who's used an old Qualcom QCP-820, QCP-1900 or QCP-1920 on the Sprint network back when it first launched can attest to this.

    However, carriers HATE using 13k QCELP because it seriously degrades channel capacity. You can't pack as many users on a channel running 13k as you can 8k. So along came EVRC:

    3. EVRC CLAIMS to be able to squeeze the quality of 13k into a stream that averages about 8kbps. IT does this by using a variable bit rate for encoding speech. With QCELP, you can be silent and still use 8kbps or 13kbps of bandwidth. With EVRC, silence uses 0kbps, and complex sounds can use up to 13kbps.

    Is it truly better? that depends on people's opinions. Some people think GSM's AMR sounds wonderful, and they hate the sound of CDMA's EVRC. Others (like myself) find AMR encoding to be way to rasssssspy ans ssssibilant, while CDMA is perfectly fine. So if you're not happy with GSM, then chances are you might like CDMA a bit better.

    I can tell you there are a few noticeable drawbacks to EVRC encoding. For one, it's designed to heavily filter out background noise. The problem is that sometimes it can't tell prolonged faint sounds (like say, soft hold music) from background noise. So after about 30 seconds or so, the hold music you might be listening to will muffle out until it turns into white noise and sounds a lot like static. For everyday conversation this isn't such a big deal though.

    There's also been talk in the past about yet another vocoder coming to CDMA alled Selective Mode Vocoder (SMV). It promises to squeeze even more capacity and quality out the network than EVRC, but that was two years ago, and it doesn't look like it's rolled out yet. Perhaps we'll see it when the transition to 1xEV-D0 or 1xEV-DV is complete.
    Last edited by scaredpoet; 12/03/2004 at 11:44 AM.
  3. #3  
    Thanks for the great explanation. I have a Cingular Treo 600 for work and a Sprint Sanyo 4500 for personal use. The voice quality is night and day, and exactly along the lines you describe. I have a convertible and can't use the GSM phone, even with the top up, because of the background noise. When on my CDMA sprint phone, though, I can talk even with the top down. While I suppose you could see noise filtering as bad in some cases, it vastly increases sound quality in my experience.

Posting Permissions