Actually, it is clipping. You are apparently driving a speaker, or perhaps a scope with the input set to 50 ohms input impedance, rather than an amplifier. Your complete circuit looks like
simulate this circuit – Schematic created using CircuitLab
You are feeding the circuit with a large signal, 1 V rms would be my guess, and this has a pk-pk of 2.8 volts. Multiplied by your gain of 10 gives about 28 V p-p, and this is clipped to about 9 V p-p. This (roughly) square wave then passes through a high-pass filter produced by the cap and speaker, and the speaker impedance is so low that the time constant is well below the period of the signal. This gives the characteristic steep rise followed by an exponential decay.
Either reduce your input level by about a factor of 10, reduce your amplifier gain, or increase your load impedance.
In comment you say you're feeding a 2k impedance, and I believe I mis-calculated the high-pass effect. 2k/0.1 uF will give a corner at about 800 Hz (time constant of 0.2 msec). This will have a noticeable effect on any audio you play through your amplifier. and it accounts for the waveforms you show, assuming you used about a 1 kHz signal. You'll need to increase your C3 from 0.1 uF to about 10 uF to get the break point down to 80 Hz, and 100 uF would be even better in terms of audio quality.