Speech transcription is at the heart of Poodll ReadAloud's automated evaluation features. Poodll uses more than one transcription system. Poodll will do its best to choose the best system automatically for the browser, activity type and passage language. But there are several options which you can use to customize this behaviour.
Passage language
This is by far the most important setting. The default setting can be altered in the Poodll ReadAloud admin settings. If not explicitly set, it will default to US English. Poodll will attempt to transcribe the users speech into text using the passage language.
Just beneath the passage language is the TTS Voice setting. This does not affect speech transcription, but should at least be a voice of the same language as the Passage language.
Line and Passage Transcriber
Poodll uses a combination of browser based recognition, 3rd party transcription services and Poodll own speech to text engine (Poodll STT). You may get more accurate, faster or more generous matching by selecting a different transcriber
Line Transcriber
This setting controls the transcriber used when doing line by line reading in "Practice" mode.
| Open STT (Strict) | This will use a standard speech to text system that does not know what the student is expected to say. It wil be less tolerant of poor pronunciation or accented speech. |
| Guided STT (Poodll) | Guided STT is more generous (ie will find fewer mistakes) because the words that appear in the spoken line are given higher probability. |
Passage Transcriber
This setting controls the transcriber used when doing entire passage reading, either in "Shadow" or "Read" mode.
| Open STT (Strict) | Guided STT is more generous (ie will find fewer mistakes) because the words that appear in the spoken passage are given higher probability. |
| Guided STT (Poodll) | This will use a standard speech to text system that does not know what the student is expected to say. It will be less tolerant of poor pronunciation or accented speech. |