The Big Blue Yeti, Soft Phones & Audio Sample Rate

A short while ago friend and telecom luminary Dave Michels contacted me about a problem he was encountering with his Blue Yeti USB microphone. While he appreciates the benefits of a headset, he prefers to not use one when there’s video involved.

Dave uses the Yeti when recording videos and participating in various UC podcasts. He’s recently started to use it with the Dialpad soft phone. That’s the service that provides his home & office phones.

The Yeti is a fine microphone for many purposes. The combination of USB convenience, handy level controls and low-latency monitoring makes it an excellent choice for podcasters. I recently wrote a blog post for ZipDX that describes its use by a professional interpreter in the UK.

In Dave’s case, when using the Yeti with Dailpad others on the call would complain that his volume was very low. So much so that he was forced to switch to his Plantronics Savi headset. They also complained that “he sounded bad.”

To solve these problems the two of us set about a quick investigation. What we found is potentially useful, so I’m sharing it here with y’all.

Testing 1,2,3: Audio Level

The Yeti is a multi-pattern condenser microphone. In his own experimentation Dave had tried the various different microphone patterns, which are selected using a knob on the back of the microphone.

The available patterns are:

• Omni-directional – Picks up sound from all directions.
• Cardioid – Emphasizes sound from the front of the microphone.
• Figure-8 – Emphasizing sound from the front & back of the microphone.
• Stereo – Uses both audio channels, split in a left & right manner.

In Dave’s case, or any podcaster really, the appropriate pattern when working alone is the Cardioid pattern. It delivers the best rejection of the ambience and reverberation of the room.

If you were sitting opposite someone at a table, and sharing one microphone, you could use the figure-8 front/back setting. That would pickup both of you equally, which could be handy.

During our test call Dave had the Yeti set to the omni pattern. That resulted in an excessive amount of room ambience and a marked drop in his voice level. Switching to the cardioid pattern improved his sound dramatically.

Sample Rate

There remained a problem that was more technical in nature. His voice stream was still garbled when he used the Dialpad soft phone. The fact that this didn’t occur when using the Plantronics headset pointed in the direction of the problem.

A headset like that is designed for voice applications. It operates at an audio sampling rate of 8 KHz or 16 KHz. In contrast, the Yeti is designed for musical applications. It defaults to a much higher sample rate of 48 KHz. Some Yeti variants are capable of 192 KHz!

Diving into the Windows sound settings for the Yeti we found that we could force it’s sample rate from then default of 48 KHz to a more modest 16 KHz. When we did this the garbling of his outbound audio stopped.

Based upon this we infer that the Dialpad soft phone can’t cope with high-sample-rate sources. That’s a bit unusual. Counterpath’s Bria for Windows is that soft phone that I most often use. It has no issue at all with 48 KHz sources.

The 16KHz sample rate would be poor for recording a vocal, but it’s absolutely fine for telephony. Some podcasters may prefer higher quality, but Dave was connecting using a frequency constrained channel, so it would not impact his work at all.