Blog Post

What does HD Voice sound like on a mobile VoIP call?

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

Here in the U.S., we may be ahead on LTE subscribers, but when it comes to voice quality on our mobile phones, we’re still lagging behind. Network operators in other countries have adopted HD Voice and wideband audio codecs to provide a better voice calling experience: Think Three UK, Telestra and T-Mobile USA’s parent company, Deutsche Telekom. In 2011, these and other carriers around the world took mobile voice calling to the next level.

I’m starting to get a taste of these voice calls, because I’m using Bria’s Android 4.0 client on my Galaxy Nexus. I previously wrote about how I’m essentially getting free voice calling through VoIP on my Galaxy Nexus with a data-only SIM card. For $30 a month, I have unlimited data and texting, so I’m using the data plan for VoIP calls and supplementing it with Wi-Fi calling in the home office.

Android(s goog) natively supports the SIP standard for VoIP, and while I can (and was) using the standard Android phone client for calls, Bria has already won me over. There are many reasons, but for now, the main one is the wide number of audio codecs supported. Instead of a “narrowband” audio codec, which limits the frequency range of a call, Bria has several “wideband” codecs (some available as an in-app purchase), bringing richer sound as well as a reduction in background noise.

To give you an idea of the difference in call quality, I recorded three different scenarios: a standard cellular to cellular call, a cellular to VoIP call over 3G, and a cellular to VoIP call over Wi-Fi. The best call experience with Bria is a true SIP to SIP call because the audio doesn’t have to be converted for a voice network during the call. I don’t have a recording of that, because I’d need two SIP accounts to call myself from one phone to another.

People I’ve called have said the voice quality is fantastic on their end. That’s a win for me, as are the essentially free calls I’m making and receiving. Once voice calls actually become data — watch for this next year with Voice over LTE, or VoLTE — I expect start to see call quality improvements, provided carriers allow for the extra bandwidth used by the wider range audio codecs.

As another example, here’s an HD Voice demo that Telestra put together to illustrate the difference:

15 Responses to “What does HD Voice sound like on a mobile VoIP call?”

  1. Comcierge Group, LLC

    I certainly understand why T-Mobile is much better when it comes to voice quality. I am sure glad that merger did not happen .. can you use the Bria Client with the Samsung galaxy4G?

  2. Todd Carothers

    Keep in mind the calls traversed the PSTN which means the sound quality downgrades to narrowband quality (although you could still hear a slight difference between cellular to cellular and cellular to HD). If you call SIP to SIP using G.722 or SILK HD the quality is fantastic – full range with little background noise.

  3. None of those calls are actually in HD Voice because, to make a HD call on a mobile, not only does the provider need to have enabled it on their network but the mobile phone also needs to support the AMR-WB codec (your Galaxy Nexus does support it, i believe, but very few other phones currently have support and none of the networks in the US support AMR-WB).

    Even then you wouldn’t get a HD call because, AFAIK, none of the mobile networks currently supporting HD Voice support it across networks.

    The only way to get HD Voice at the moment is on two mobiles that support AMR-WB on the same, HD enabled, network or by using VoIP end-to-end.

    The two calls that included a SIP endpoint did sound slightly better, but i suspect that’s because the SIP end of the call was encoded in G.711 and so had to be transcoded fewer times (mobile to mobile would be GSM to G.711 to GSM, whereas SIP would be G.711 to GSM).

    To truly hear a HD call point your SIP client to [email protected] while using the G.722 codec. The difference is really astounding and is set to get even more pronounced once the IETF’s new Opus codec starts to appear on devices.

  4. in the youtube video with the Telstra engineer, when he mentions that some people say “that doesn’t sound like me”, i would add that when you speak, you hear yourself through bone conduction and resonance too. your voice frequencies literally passing through your body from vocal chords to ears. not just the vibrating air from your mouth and later going into your ear from outside.

    this means that you speak you can hear, and feel, your lower frequencies through your own body, while the listener elsewhere completely misses out on those.

  5. I’ve been thinking about this. Do we still need voice on smart phones our can we migrate to using voip over a data plan? A few questions.
    * How reliable is the voip compared to using regular voice. Does it work well when you’re in the car, different areas, etc?
    * Can you start a call on wifi while at home and have it move over to 3G when you move out of range?

    T-Mobile has a pre-paid $30/month with 5Gig 4G and “unlimited” 3G. I’m tempted to use it, bring my own phone, and no more carrier dictating what device I use, what I put on it, etc.

  6. Luke Freiler

    I may be in the minority, and I assume flash/youtube aren’t the best way to demonstrate the quality of this tech, but they all sound very poor to me. I’m used to using Skype daily, on a variety of mics, which sounds brilliant – and compared, the best of these sounds about 5 years back.

    • True, this isn’t the best way to demo the sound quality. I should note that Bria can use the same SILK codec that was created by and used for Skype. For optimal results both callers would need to be using VoIP for the reason mentioned in the post.

  7. ishekhar

    Intriguing article. Definitely got me interested in researching more.
    But to be honest, I didn’t’ really find any noticeable difference in the 3 sample sound bite that you recorded. If you spoke the exact same sentence all three times, I probably couldn’t’ have guessed which one is from what setup :)