Note: after some calculation, I consider this impossible. Dumping for future explorers and in case I’m wrong.
While CC1101 doesn’t support analog radio, it might be possible to receive NFM on UHF - some portable radios live there. To do so, make a few AM receptions on adjacent frequencies, discarding data, but retaining RSSI. In fact, I got this idea playing with my walkie-talkie and figuring out that Spectrum Analyzer spectrogram looks almost normal.
After that, we have a FM spectrum snapshot and some DSP will get us an audio sample.
So, how bad will it be? Critical part is acquiring many samples fast enough. Retune time when hopping between predefined frequencies is 75 us (reference: section 28.2.2 of datasheet). Assuming we read batches of 100 subsamples, getting slightly less than 7-bit audio (with FM, more frequency resolution on input = more amplitude resolution on output; 7-bit speech is OK, and we can’t get much more due to our speaker), it takes 7.5 ms for one sound sample. Reciprocal: audio rate will be 133 Hz, leaving voice tones of less than 60 Hz per Shannon-Nyquist. Normal voice is around 500 Hz.
Conclusion: Flipper’s hardware probably won’t cut it.
If you find any mistakes, please tell me.