Scope the output of the Tl494. Hissing usually means audible piezoelectric effect off your caps. I have seen a TL494 'shift' its PWM output oscillation frequency downward in binary steps ( by implementing additional dead time) to control the target current load, then jumps back to the original base frequency when the load is corrected, causing a modulated PWM pulse train which 'sounds' different because of the caps 'buzzing'. If you can detect such unexpected PWM frequency shifting over a few PWM cycles you need to look closely at the error amp circuitry being faulty. This type of unwanted modulation increases heat in the output driver circuitry. Sometimes you can correct this by adding capacitance to the error amp input to act as a simple low pass filter thereby preventing over corrections. Reducing the error amps feedback (amplification) can help a bit as well. Adding additional decoupling close to the TL494 supply pins is also a good idea.
I have found it preferable to use an external, low slew, dual opamp to amplify a precision current sense resistor's V and a difference amp to directly control the DTC input as a means of reliable control when designing with the TL494. I use its internal error amps as thermistor based current limiters.