KIG63 was off due to an error on my part trying to get a better signal (reducing gain), but that was corrected. If you manually listen to the Petoskey or Hagerstown streams (hollow star) and compare it to mine, which is better? I'm not sure bit rate or sampling makes that much of a difference for this type of broadcast.
Why KIG63... obviously... no contest, no argument. It sounds better than many, in fact. It's been "good" since it was added...
...and the referenced feeds have also been 'good', or at least much better. Get a good signal to them, then what?
What some may question are the 'pre-qualifier stream parameters' --- what long hours, experimentation, and testing have determined presents the product best to the listener, overall, and in various 'situations' ... those are "Is the Audio In Both Channels (Or Full Mono)?" Are both the digital audio stream's BitRate and Sampling frequency sufficient for best quality under a wide range of conditions?
Also keep in mind that the analysis generating 'stars' reflect ONLY what was sensed at the time of any analysis, and could change in moments, or the next morning... Static / noise
by itself will NOT kill a Hollow Star....but can 'demote' a 'Solid' to 'Hollow'... The analysis is automatic, not manual... and under refinement, even as this is typed...
We also felt originally as mentioned, regarding the 'rates'... We didn't expect the need for higher 'bit rate' and 'sampling rate' when we began this project about a year ago. And in fact, we need to update some of the information on the website.
We first believed a bitrate of 8Kb/s and a sample rate (frequency) of 8KHz, (and then later - 16KHz) should be adequate. In fact, may work well on pure audio BW of 3750 Hz, in most cases...
But,
4KHz is NOT the max freq being streamed on virtually all sources....! At the very beginning it may be received by an FM radio not operating as 'narrow band', and processed through a 15KHz BW FM circuit, and then sent to a device with an encoder installed, which converts to digital MP3, with some sampling frequency and bit rate, sent through various connections to a device which may be operating at some different "personal settings".... including volume level setting.
Most streams, then, include noise, hiss, hum,at varying levels, etc to 15KHz or higher>.. Many have some clipping, excessive input volume, etc... None of which existed in the original NWR audio before it left the transmitter. We were wrong, overall.
Long hours were spent. This established guidelines across many 'situations' and 'encoders' and 'reception' variables. Today those parameters have been locked in at 32Kb/s (Bit Rate) for smoothest audio levels and 22,050HZ Sampling Frequency minimum for best frequency reproduction with minimum aliasing. In fact, we could suggest the CD standard 44.1KHz sampling, again because of the extended audio bandwidth of most audio amplifiers connected to the encoders....and don't ignore the wider bandwidth on the 'listener's' end, as the 'digital mp3' is recoded to Analog.
One might note, especially to the 'louder streams'.. at times the audio may 'pop' ('thump') or 'blank' for a moment...even 'appear' to 'lock' or 'skip' ... that can be due to a low bitrate, especially when coupled with a low sample frequency and higher levels.
We've been evaluating this for months. Throw in noise, low or excessive levels, a single channel sent, especially when encoded as 'stereo', weak reception, then the Bit Rate and Sampling frequency baselines became apparent. These were not set arbitrarily.
So besides Overall Average Quality, and Stream availability, which qualify for a stream star,
the 'pre-qualifiers' areStream uses both channels if Stereo Encoded, or Full Mono Mode ( with both channels)
Stream bit rate of 32Kb/s or more
Sampling rate (frequency) of 22.050 KHz/sec or more
Those three directly affect the Overalls in
many cases, are easily reconciled, and present the stream better to the listener... regardless of any failures in the 'graded' 6 point analysis.
