The core of TTS family products is a TTS engine with the set of our own original voices. The TTS product always consists of two parts - TTS engine and the voice. Versions of those parts used in the product determines the quality of the resulting synthetic voice and performance of the product in computing power.
Currently operates TTS engines of versions 2.6, 2.9 and 2.10. The quality of synthesis for all versions of the TTS engine is different. The version 2.6 is fast but of lower quality of voice. Versions 2.9 and 2.10 are comparable of higher quality but the new 2.10 version is much faster. Likewise, even with the voices. The quality of the voices is determined by the intended scope and quality of original recordings and the degree of the subsequent "post-processing" and "cleaning". Simply identify the quality of the voices by the number of stars - the more the better quality synthetic voice.
Czech voices
| voice name [lang] | version | download | quality |
| Alena [CZ] | 2.6 |  |
 |
| Alena [CZ] | 2.10 | |
|
| Iva [CZ] | 2.10 | |
|
| Jan [CZ] | 2.6 | |
|
| Jan [CZ] | 2.10 | |
|
| Radka [CZ] | 2.10 | |
|
| Tomáš [CZ] | 2.6 | |
|
Slovak voices
| voice name [lang] | version | download | quality |
| Melánie [SK] | 2.6 | |
|
| Melánie [SK] | 2.10 | |
|
English voices
These two voices are copyrighted by CereProc Ltd. See more information about this company at their website www.cereproc.com.
| voice name [lang] | version | download | quality |
| Cereproc Sarah [EN] | |
||
| Cereproc William [EN] | |
|
Note: the stars at names evaluate simply the quality of synthetic voice
All the voices you can try at our online demo.
The software is platform independent and supports the following platforms:
- Intel, Windows (2K, XP, 2003, Vista, 7), 32 bits
- Intel, Windows (2003, Vista, 7), 64 bits - 32 bit library
- Linux, 32 bits
- Linux, 64 bits
Custom text synthesis
Based on your order, we can professionally convert your text into audio files. This service is also spot-checking of converted speech and the preparation of the text entered. The text entered first semi-examine and possibly unfamiliar abbreviations or inappropriate parts of the text is adjusted so that it can be read better. The resulting text is converted to MP3 or to other desired format.
The service is suitable for occasional use such as TTS without having to purchase expensive licenses and training personnel TTS to use the new program. It is useful for tasks sound presentations, documentary films, the synthesis of the texts on Web pages, generate a single sound for audio recordings of your program, etc. The price is determined as 1 page of text, where a larger number of pages we would like to offer a volume discount.
Online TTS
Synthesis of texts through a proprietary asynchronous XML-RPC/http interface. Suitable for telephone systems or Web applications without any requirements to generate in real time. Can easily take advantage of various programming languages and frameworks, such as PHP, Python, C / C + +, Javascript, Flash etc. Licensing Online TTS is the amount of synthesized data. For offer, please contact us.
TTS Server
Partner - Suitable for tight integration with partner products through a proprietary low-level API (Linux, Windows). For the closest possible integration into your product.
Radio system with TTS
Specialized TTS module can be very conveniently used to broadcast messages in public spaces or for the calling, personalized reporting system. Form of implementation is dependent on technology, customers - please contact us for details. Licensing is dependent on the number of locations where the system operates.
TTS for mobile and embedded devices
For devices with low power we offer SpeechTech TTS version 2.6, which is a performance suitable for such devices. TTS only provide partners for integration into their own products. The product is licensed through the payment for each device or as a revenue-sharing. We are able to provide a library for mobile phone Apple iPhone.
TTS voice on demand
On the customer’s demand we can create a TTS with a special voice, e.g. a voice of selected spokesperson of the organization - provided that the speaker records a speech database - it takes few weeks dayly several hours. Dealt with by the project - the time of creation and delivery of voice is about 6 month.
It is also possible to increase the naturalness of computer TTS using so called a domain-oriented TTS, for which it should take frequent recordings of sentences and phrases to be present during the operation. This option allows you to escalate the naturalness of speech normally delivered by the voice of the customer's specific use.
Licensing
According to the method application we offer:
- license per a channel of IVR system
- license for the geographic location of radio systems
- license to install the TTS library without the possibility of further dissemination of data generated
- license to distribute the generated audio data counted by data amount
Properties
- The input may be used an unformated text, to benefit all the features of SpeechTech TTS we recomend use the standard W3C SSML (Speech Synthesis Markup Language).
- Supported platforms:
- Linux and Windows (XP/server 2003/7 PRO/server 2008), 32bit i 64bit architecture.
- Simple API for easy integration with custom code from: C, C++, C# (Visual Studio), Java (JNative), Python (ctypes)
and others.
- Distribution in form of DLL / SO libraries.
- Possibility to compile for embedded platforms (ask us).
- Supported MRCP in frame of product SpeechTech MRCP server