Sandbook: Natural Voice Reader

I really like that the voices are realistic and they actually work. I like the simplification of this app just take a picture and it scans it. However, I don’t like that every uploaded photo the audio states “page 1” “page 2” etc. for starters…. those are not the actual page numbers in our book because we have to constantly be scanning new pages within new chapters of the book. It’s distracting from the text and the flow of the audio. Also, even if got scan a photo of the page perfectly somehow the Audio states a word that is sooo contrary to what the actual word is. A lot of mistakes on the audio reading simple words. Additionally, I don’t like the limit of 24 pages of scanning. That doesn’t cover the entire chapter of a book. Also, I don’t like how I cannot name the bundle of pages I’ve already scanned. By far, The worse thing is the fact there is no background audio play back. Meaning if my phone gets on “locked screen” the audio will immediately stop. The developer thought adding ONE PAGE audio playback option was enough? It takes less than 5 mins to read a page… you want me to select the button to play back the next page every single time through out the book? That is ridiculous. Amongst other issues. I look forward to seeing this app updated to improve on these features and then to become the #1 app of its kind but right now- these are my issues with it.

The voice itself is great. It sounds very human. But it struggles with pronouncing certain things and where to pause in sentences and phrases. The PDF I fed it had the word clothing split between two lines with a between CLOTH on first line and ING on second line. It pronounced it as two words instead of one. Another example is the words great apes. It pronounced it “gray-it ai-ips.” The word laid was pronounced “lay-id.” It also pauses at the end of every line for as long as at the end of every song entrance, making the reading choppy. The funniest was when humans was split between lines as hu-mans and it pronounced it “who” pause “mans.” Would love using the voice, if the reading flowed and pronunciation improved!

The speech started off sounding so surprisingly *normal*/acceptable (which is above par for local iOS TTS). And then it reached the many words in the source text that include -eight (eight, weight) instantly revealed that all voice/region variants apparently come from one distinctively unusual/foreign source. It would be great if we could train/adopt new voices based on voice recordings (voice cloning). Continuing to avoid using this for now.