TTS Tools and Language Instruction

TTS (Text-To-Speech) tools are computer programs whose function is to read any text aloud. Such tools draw on a database of recorded sounds or words which they use in processing the written text. When the database is comprised of words or phrases, the tool is only capable of reading a text that contains these particular words or phrases, which requires an environment where very little variance is expected. For example, an automated system announcing the time of the day in the format "the time is twelve thirty PM" only needs the words for the numbers one through fifty nine, AM and PM, and the phrase "the time is". When the database is comprised of individual sounds, however, the tool is in principle capable of reading aloud any given text in that it can account for any given combination of sounds. When such a tool is also programmed to incorporate prosodic elements such as intonation and rhythm, the closeness to natural human speech can make the result quite appealing. TTS tools, often referred to as speech synthesizers, are widely used in a variety of telecommunication services where information is provided via telephone messages. They are also common in announcements pertaining to flight information, weather, and the like. Such tools are of great value as assistive devices in the service of people with disabilities (for example reading aloud a book or a web page for a visually-impaired person, or reading aloud a text typed by a person with a voice disorder). If they are sophisticated enough to produce speech that is close to that which would be produced by a human, they can also be of great value in the context of computer-assisted language instruction.

Since a high degree of naturalness is mandatory when TTS technology is applied to language instruction, tools developed for such purposes must be language specific. They must rely on the sound inventory of native speakers and on a thorough analysis of native speech patterns. A variety of high-quality voices to choose from can enrich the learning environment in that the learner can hear the same text spoken by a man, a woman, persons of different ages, and so on.

This page focuses on the use of internet-based TTS tools in foreign language instruction. We provide here a number of ideas for utilizing such tools, and then identify a number of sites and tools that language instructors and students can use while implementing these ideas.

Most language-instruction web sites incorporate sound files that correspond to the curricular materials. The use of TTS tools in conjunction with a given curriculum provides the additional benefit of allowing students to control the segments that they want to hear, and work with an unlimited choice of structures and vocabularies that they can type and listen to at will. We find internet-based tools particularly suitable for that purpose because they are easily accessible and often free or inexpensive. For the most part, if such tools require a download, it is generally simple and straight forward.

We provide here, then, links to and short descriptions of TTS tools capable of handling a variety of languages. The majority of these tools are commercial products, some of which are offered to developers for incorporation in their own programs. Our list is based on the availability of online interactive demos or trial versions that instructors and students can use, and our understanding is that such use is for educational purposes and that users may want to test a product in preparation for purchasing its full version should they find it useful. Instructors may want to note that Microsoft Agent, when configured appropriately with the desired language components, can perform TTS tasks in a number of languages. While it is free of charge, it requires a complex download, and is also lacking in the area of naturalness. Thus we do not recommend it here. The degree of naturalness varies greatly between tools and between voices/languages within the same tool even in the list that we provide, and instructors may want to examine the tools carefully before they use them with their students. While we are aware that much is left to be desired, our main goal is to create active interest in the combination of language instruction and TTS tools, and thus create a demand for better and more affordable speech synthesizers.

TTS tools can be used to reinforce correct pronunciation in a foreign language. Language students may benefit from developing the habit of hearing selected phrases from each day's assignment spoken by a native speaker. Thus, they are more likely to say correctly what they are learning to understand and write. Lacking such input, students often develop their own personalized systems of pronunciation, which could differ markedly from that of a native speaker. It could subsequently become a lot more difficult to "unlearn" poor pronunciation and relearn correct speech patterns.

In addition to pronunciation, listening comprehension can also be improved with listening becoming a habit, and correct spelling patterns are likely to develop since incorrect spelling will produce a vocal rendition that students would recognize as incorrect.

TTS tools can also help students assimilate more quickly and effectively new vocabulary and structures. While this is especially true for auditory learners, better learning is likely to take place when the auditory element works in conjunction with the visual one. Our assumption is that TTS tools can assist in building confidence and reinforcing good learning habits, and for this reason we recommend that first-year students get into the habit of working with such tools as soon as they begin their language study.

Basic ideas for utilizing TTS tools in language training:

• Students type a word, a phrase, or a paragraph and listen to what they have written in a variety of native-speaker voices. The input can be a segment from a given text that students copy and paste, or a text that the students type on their own.
• Students type a response to a question or a cue, listen to it, and then compare it with a parallel sound file to identify structural or spelling errors. This type of exercise is appropriate primarily for situations in which only one correct response exists.
• Students send short e-mail messages to one another, and the person on the receiving end hears the message read aloud.
• Students try to approximate native speech in a target language using a TTS tool that is developed for English. While doing that they can identify differences in the sound systems and intonation patterns of the two languages. Such exercises can be very amusing, especially when they are done in groups.

Check our list of Web-based TTS tools