Several problems with the circuit. First off, LM741 has common mode range of up to +/- 3V less than power supply so in other words your usable voltage range is only +/- 2V. +/- 15V voltage range is recommended. You have +1V DC bias on the input so straight away the poor amplifier is trying to push -5V to the output, which it's incapable of reaching. Usable output voltage is also 3v less than rails so +/- 2V also.
So in order to get any kind of reasonable operation, add reasonably large capacitor in series with the audio signal. Reasonable in this context would be e.g. 47uF (plus towards audio source!) which will affect amplification by about 6% in the low frequency range if you use 1k/5k resistors. Increase those resistors by factor of 10 and this sinks to <1%.
You also need to add power stage because that opamp is incapable of driving the speaker. To do so, you need to change the circuit to non-inverting because you need to get a feedback from the output stage.
Here's a circuit that will work with a bit of modifications:
What you need to change is obviously replacing +6V with +5V. VR1 and speaker are connected to actual GND and other ground points in that schematic go to -5V. You probably don't want that 100k / 5k voltage divider either, for 5x amplification you'd want 39k and 10k (4.9x).
Finally, you can drop the capacitors from speaker and feedback resistor.
As a parting shot, I'd either get +/- 12V power supply or if that's not practical, replace LM741 with for example TLV272 (dual) or TLV271 (single) which is a rail-to-rail output but can't do more than +/-8V (or 0 to 16V).
You could also hook the TLV272 to a 12 volt supply and use the circuit as-is.