SNR is often expressed in dB, and so the value is negative whenever the signal level is less than the noise. Thus, it appears likely the SNR values you extracted are expressed in dB. In order to convert to an actual ratio (of all positive numbers) simply transform using unitless_SNR = 10 ^ (SNR_in_dB / 10).
It's not clear what you want to do. Do you want to normalize everything in the range from 0 to 1? But are the negative values for SNR expected? I guess not.
Consider the range from the lesser to the higher value. Shift this range to start from zero and then divide all elements in the array by the highest SNR value. Your array is normalized.
SNR is often expressed in dB, and so the value is negative whenever the signal level is less than the noise. Thus, it appears likely the SNR values you extracted are expressed in dB. In order to convert to an actual ratio (of all positive numbers) simply transform using unitless_SNR = 10 ^ (SNR_in_dB / 10).
The formula stated by Charles Mesnier is to convert dB to Voltage or Bits (computer Lanuguage). You can google for conversions for dB you can get the reference.
As Mr. Narne recommends, just use Google (or any search engine, or even the index of almost any engineering-level math text) to learn about dB conversions. I usually open the Wikipedia entries that result from my own searches first, and in this case, https://en.wikipedia.org/wiki/Signal-to-noise_ratio.
The SNR_dB = 10 * log_10(unitless_SNR) equation is provided as the third equation in Section 1.1 - Decibels. While the Wikipedia entry does not explicitly show it, the conversion from a ratio expressed in decibels back to the unitless ratio is just a simple algebraic rearrangement:
SNR_dB = 10 * log_10(unitless_SNR) [ the conversion to dB equation]
SNR_dB / 10 = log_10(unitless_SNR) [ divide both sides by 10 ]
10 ^ (SNR_dB / 10) = unitless_SNR [ inverse log_10 of both sides ]
unitless_SNR = 10 ^ (SNR_db / 10) [ the conversion from dB equation ] ::