Main Site Documentation

A new text-to-speech module/IC


#1

Hi, --:

Along the course od developing a security system, I scouted a text-to-speech IC XFS5051CE. I think it might interest folks in this forum so I’ll post it. The best part of this IC is, you can burn as many wav or pcm files as you want into a SPI flash, then play those files just like synthesizing speech from text… The following is text I sent through UART. Bolded words are the names of hard-burnt wav files.

The synthesized audio can be downloaded from my dropbox:
https://dl.dropboxusercontent.com/u/23053956/XFS5051-TTS.wma

ATT and Festival are audio files of something read by either ATT Natural Voices Lab or The Festival Speech Synthesis System . Honestly chip/hardware based TTS is still not as good as software based solution but for us embedded people often times we simply don’t have access to internet. So I have no complains.

This IC handles volume, pitch and speed, but I don’t have the chance to play with those features yet.

My plan was to use S1V30120 from Epson but the minimum order of 600 just scared me off and Joe’s EMIC2 seems quite expensive for what it can do. So I looked a little bit harder and found this IC.

1K unit price of this IC runs at $15, kind of expensive too but given its capabilities to handle audio playback(either from SPI flash or from MCU through UART), I think it is still good choice particularly for .Net Micro Framework. NTMF so far doesn’t support I2S. Audio play needs to be handled by external IC or native code. Combining TTS and audio playback into one IC greatly simplify system design when both features are needed.


#2

Gus, looks like another module to make. :slight_smile:


#3

I agree EMIC2 is an expensive platform (but I have one :slight_smile: ). The Epson chip certainly didn’t seem “friendly”, even though I didn’t even get to look at MOQ’s. But as far as an all-in-1 module I found it reasonably easy to get to terms with (I hate the reliance on just serial feedback rather than a “busy” output pin though, that’s non-trivial to deal with, so Michael if you seriously consider this, and the chip is capable, please put a busy pin on the module !!!)


#4

Not bad! Although, sounds not very good on the phone for some reason.

So, where can you get a sample?


#5

I2S support would be a great feature for the community firmware. If I understand correctly, it’s similar to SPI (uses the same MCU peripheral even)?


#6

Maybe it’s the geek in me, but that has shades of HAL to me? :slight_smile:

When can we get one?


#7

Thanks for response, folks.

@ Michael, Gus: This chip is really a powerhouse. It can even encode audio if a microphone is attached. It also can communicate with MCU through UART/I2C/SPI. You guys should seriously consider making a gadgeteering module out of it.

@ Brett: This chip has a busy/ready pin. It drops to low when it finishes synthesis.

@ Dave McLaughlin: I bought 10 chips, five of them are running in my security system. The other 5 are just laying around. If you are interested, I can send one board to you. Just pay me BOM which is about $30 plus shipping :slight_smile:

@ Architect: you are right. The quality of recording is not good enough even to my own ears. I am not sure it is because my creepy line-in cable or $60 motherboard or my layout. I’m still working on it.


#8

You’re on!!

Drop me a line with details. I can pay via Paypal if you have that?

dave at axoninstruments.biz


#9

I think this deserves further explanation :smiley:


#10

I wonder if he has them talking to each other?


#11

Well, I didn’t make them talk to each other because I don’t have a voice recognition module if there are any. I do sometimes make them talk non-sense like reading speeches of certain kind of people when I need some noise. :slight_smile:

@ Dave McLaughlin: I am writing up a brief instruction, when that is done I’ll email you.