As a student, I should be able to listen to "pyramid"-style tongue-twisters, so that I can learn difficult language sentences.

[jrovu]

Requirements / Concerns

• Number of columns / length of sentences (shorter sentences should not fill space with pause)
• compound words vs single words have different word sound
• pause = .5 second
• No text space for Chinese, but other language probably. read numbering column like others we've done start with whole sentence (play complete sentence first for the listener so they know what they're practicing)
• although I don't think we need this for every sentence, it still would be nice to have. The only reason "not" to have it is because of export time, not because it wouldn't be useful. In other words, if later we can shorten export time, we should have this option for every sentence, not just tongue twisters
• 80% speed

Example: Pause + I'm going to the bank + Pause + I'm + Pause + I'm going + Pause X2 + I'm going to + Pause X3 + I'm going to the + Pause X 4 + I'm going to the bank + Pause X 5 + pause (just to end it)

Phrase Pyramid Audio - From one phrase, it creates Complete Phrase + Pause + Word1 + Pause + Word1 + Word 2 + PauseX2 + Word1 + Word 2 + Word 3 + PauseX3 + Word1 + Word 2 + Word 3 + Word 4 + PauseX4. As an end user, I'd like a listen & repeat MP3 that starts with the first word + a pause, and keeps adding the next word to "hold my hand" through a sentence.

[jrovu]

FYI @ctparadise I moved these comments from the XLS and created this feature issue.

[jrovu]

Made good progress on this today.

Accomplishments:

• The program is able to (1) read the CSV containing the phrases, (2) Create "pyramid" text from the parts (e.g. "I", "I am", "I am going"), (3) Create audio files for each pyramid

Next steps:

• Combine the pyramid files, along with appropriate padding

[jrovu]

This is the way to run the program in "pyramid" mode.

./tts.py -f pyramid_chinese.csv --foreign_voice Zhiyu --english_voice_engine neural --mode pyramid --speed 80

If you encounter a problem, run it in "verbose" mode, like: ./tts.py -f pyramid_chinese.csv --foreign_voice Zhiyu --english_voice_engine neural -v --mode pyramid --speed 80