I didn't write a pipeline or anything complicated. I just used google console. I used a recent en-US-Journey-O voice as it seems better suitable for narratives. I used synthesize long audio APIs. These support upto 1 million characters (currently). I had to chunk the book into smaller parts though since that was it was easier to edit things if it didn't sound right to the ear (from pauses perspective). I think it did a reasonable job. At few points the narration seems a bit out of place and not quite right so to speak but this turned out to be quite low effort for a relatively layman.
The cloud usage cost was trivial (tenth of a dollar total).
No comments yet.
