Compress audio summaries with OGG or similar

[wchargin]

Migrated from tensorflow/tensorflow#4207.

@carlthome says:

When logging an AudioSummary every couple of epochs, the logs grow quickly (as in several gigabytes), and checking TensorBoard remotely (as in not on the same LAN) also becomes very slow. Would it be possible to have builtin support for Ogg compression or an equivalent lossy format? There are multiple use cases where this would be beneficial.