Improve performance

[maxbbraun]

Some ideas from section 4.5 of the paper and this discussion:

• [ ] Use shorter chunks to reduce latency. (Try overlapping chunks by setting options.prefix to the transcription of the overlap.)
• [ ] Try beam search instead of greedy decoding via options.beam_size = 5. (Also look at patience and temperature options.)
• [ ] Try previous-text conditioning by setting options.prompt to the previous chunk's transcription.