Help, Notes, & Troubleshooting

Please note that iSpraak is functional but still in development. Below are some tips and information to help you make the best speaking activities possible. You can e-mail any questions or report odd behavior to the developer at

Selecting Texts: Note that it is recommended that you focus on short contextualized texts. A sentence like "John picks up his mail every afternoon" is more likely to be transcribed correctly than a series of unconnected words. If your objective is to work on minimal pairs, it is also better to put these words in short contexts. For example: Paul put the disk in his computer on his desk. This is better than students just alternating the words disk and desk without any context. Also, always remember to test your activities yourself before assigning them! You may find some words (i.e. uncommon proper nouns) won't be transcribed as anticipated.

Student Scoring: Keep in mind that no speech-to-text technology is 100% accurate, even for native speakers. Because of this, consider giving your students a target goal of 80 to 90%, depending on their level and the difficulty of the sentence. iSpraak is a tool to encourage speaking practice, and is not well-suited for high-stakes evaluation.

Lengthy texts: You may find that student submissions of lengthy texts do not save properly. It is recommended that the exercises you create focus on shorter texts if you experience this problem. Try limiting your text to fewer than 90 words. Remember that short texts in a clear and simple context work best!

Text-to-speech: This is not yet available for all languages. Additionally, please note that lengthy texts of some languages will cut out after 15 seconds of text-to-speech synthesis. If you must use a lengthy text, you can upload an MP3 with a recorded prompt rather than rely on the synthesis.

Browser Support: iSpraak was designed to work in Google Chrome on Mac OSX, Chromebooks, and Windows 7 or higher. No development testing has been done on mobile devices.

Homonyms: Beware phonetically ambiguous statements and context-free homonyms (French: Ils marchent vs. Il marche). If you are unsure how the API will transcribe your speech (Brian or Bryan, Luc or Luke), it is a good idea to test the activity first to ensure the transcription is what you anticipated. You can adjust if necessary by creating a new activity.

Punctuation: iSpraak attempts to remove all punctuation before comparing the model text with the student submission. While the automated score is calculated without punctuation, you may notice that some words appear in the section because punctuation was missed. If you want to avoid this, you can create activities without punctuation, or you can tell your students to clearly say things like "question mark" or "period" in the target language. This will put punctuation into their submission. As a general rule of thumb, limit punctuation as much as possible. This amazing resource has the most common words and phrases in hundreds of languages. If you find, however, that a word in your own exercise is not listed in, you can create a free account on the website and record it yourself. It only takes a minute to register and start recording.

Segmentation: Scriptio continua is a feature of some languages that lack spaces or other marks between one word and another. For these languages, iSpraak breaks down words and sentences into individual unicode characters. You will notice, for example, that the automated feedback for Japanese and Chinese does not consist of a series of missed words, but rather individual letters, phonemes, or characters. The Chinese word for house "房子", if partially missed, may be presented as feedback as only "子", meaning "son".

Transliteration: There is some transliteration support for non-Latin script languages, supported by; For Chinese, for example, you will see a pinyin icon near the speech synthesis buttons that will provide a bubble overlay with transliterated text. In other languages, such as Russian, this button may appear as a small "ABC" icon. Contact us with inquiries about adding additional transliteration support at

Mobile Devices: Some mobile devices will override the language input setting and student output will appear as transliterated L1 text rather than the target language. A documented example of this occurred with the target text "You can count on me" being submitted as the phonetic approximation of the Korean characters "유캔카운트온미" (yukaenkaunteuonmi). This appears to occur with select mobile devices, which are not yet officially supported by the software.

Microphone Blocked: Please first make sure you have not inadvertently blocked the website from accessing the microphone. You can see a list of blacklisted websites by navigating here in your Chrome browser: chrome://settings/content/microphone
If you see iSpraak in the block list, be sure to click the delete trash can icon. You will then be prompted again when you visit the website.

Accessing Grades: From the Grades overview, you can click "Login" to see all of the activities associated with your e-mail address. If you do not see this option, you can re-enable it by clicking on the instructor link from any recent iSpraak e-mail. You can also authenticate here.

Activity Stats: To check out a list of the most commonly missed words for an activity, click the "Analyze Error Frequency" link on the grades screen. This will give you a visual representation of the four most commonly missed words, as well as an alphabetic listing of missed words and frequency. If you see an error on this screen, please contact the developer at

LMS Integration: There is a basic LMS integration via LTI 1.1 so that institutions can manage iSpraak activities through BlackBoard/Moodle/Canvas/etc. For instructions on setting up this integration, please contact . Required LTI data include role, student name, e-mail, course title and course ID number. No other sensitive student info is stored on the iSpraak server and student grades are accessible only by the instructor or student. Gradebook integration is available for most systems, but is limited to complete/incomplete postbacks.

TTS Credits: Your institutional quota of TTS credits is reduced for each word of premium text to speech that is actually synthesized. Credit is only used for the initial synthesis of an activity, and not for subsequent re-plays of the file. Please note that a ten-word iSpraak activity may actually use up to 30 credits-- one credit per word, per voice (male, female, slow). If the premium icons are not clicked by a student or instructor, the credits are not deducted. A running total of remaining institutional credits is available to the individual who requested your institutional license. Credits are only calculated per activity created, not per student submission. To view your used TTS credits, click on the "List All" link when authenticated in the program. Once credits are exhausted, additional credits can be purchased, or you can simply continue using iSpraak for the duration of your license without the ability of generating new TTS voice files. Speech recognition and file uploads will continue to work.