Speechling White Paper
Speechling was designed for one purpose: to help anyone learn to speak a language on the Internet.
Reason #1: We have a real voice you can imitate.
As of May 2017, our release, no site currently provides:
- High quality audio of real people saying real-life phrases
- Both female/male audio
We do both.
The vast majority of sites currently provide text-to-speech transcriptions of sentences and phrases. While that would be okay for practicing vocabulary and grammar, nobody wants to speak like a machine.
Speechling aims to provide the highest quality audio in crisp, un-sampled, un-compressed WAV bitstreams. All of the audio is native to our platform, which means our users don't have to navigate to YouTube or SoundCloud.
We think that having both male/female audio is essential to our users. Many languages, such as Chinese, have very different mannerisms between genders. Imitating the opposite gender often leads to embarrassment and bad speech habits.
How we select our recording staff
We carefully vet our recording/translation staff before delivering any content to our users. Each member of Speechling staff is paid above market rate for their exceptional voice acting talent. As an example, for a typical recording session:
100 people apply for the position.
50 are selected for being of acceptable native quality.
15 are selected for having the best word choice for translation.
5 are selected for rhythmic quality and enunciation.
2 are finally selected for production quality recordings.
Our studio in San Francisco contains bleeding-edge equipment, and our quality assurance team carefully inspects each audio for quality defects.
Reason #2: We give you personal feedback.
Speechling provides audio feedback to every one of our students on a daily basis. We will imitate your mistakes in an audio file, and then say the right version, side by side. A short text summary of what you need to work on is provided daily as well.
Users only submit what they are comfortable with, and can easy record/play back their own voice as many times as they like in one sitting. This encourages practicing and repetition.
A key part of learning any language is the personal feedback. Are you rolling your R's in Spanish? Are you struggling with a guttural R sound or the nasal sounds in French? What about the tones in Chinese?
However, working on the speaking is notoriously difficult. Authentic content is never paired with feedback or recording devices. Who is listening when you're singing along to an authentic song?
I dread the "pre-conversational" phase of learning a language personally. While reaching out to friends, I feel like they're doing me a favor while I stumble through the sentences. Even when using a tutoring service, I often felt ashamed for being at such a low level, or for spending so much money on correcting such basic mistakes.
Speechling offers a no judgment zone. Only recordings that users are comfortable with are shared to our coaches. Feedback is given privately and not visible to other users. This balances the "in-your-face" advantages of private tutoring with the "at your own pace" independent language learning platforms.
Reason #3: We're here with you through your journey.
We're a big believer in documenting the language learning journey. The journey to fluency can last years and mastering a language can last a lifetime. Day by day, it may seem like progress is minimal and the road is a bit too long to handle.
All it takes is a quick listen to how you sounded even a few months ago, for you to see how far you've come.
A key feature of Speechling is keeping that journal for our users. From the moment you start using the site, we journal your audio recordings and the feedback recorded.
We're all about using the latest research in education to give the best experience for our users. Some of this stuff exists on other language learning platforms, sure, but we're putting it all in the same place for you to enjoy together.
Given enough time, our brains eventually forget something that we learn. Spaced repetition tries to fight that process of forgetting by reintroducing the learned material at the right frequency.
Built into the ordering of our sentences is a uniquely designed spaced repetition system. Instead of repeating the sentences directly, the content incorporates it on the phrase level. Specific common groupings of words (e.g. "would be able to") are carefully crafted to repeat throughout the content at an optimal frequency.
We draw inspiration from Bloom's taxonomy. In it, mere "reception" of knowledge is one of the worst and lowest level ways to learn. Lectures and vocabulary lists are one of the worst ways of delivering content.
We instead believe in learning from a wide variety of exemplary sentences that are commonly said in each language. Each step of the way, students are encouraged to use what they know in their native language and internalize the translations.
Instead of being taught grammar and vocabulary explicitly, through repeating the right sentences, students are encouraged to discover instead of merely receive. It is how children learn a language, and it is our preferred method too.
Speechling is a website focused on the spoken language. The corollary is being able to listen effectively. As such, our feedback is often centered around correcting the bare bones of the language at a granular scale.
We are prepared to traverse the entire hierarchy of language for our users:
- Basic sounds of the language on the syllable-level.
- Rhythmic pattern of sentences and inflection.
- Pronunciation of individual words.
- Pronunciation of entire phrases.
- Pronunciation of an entire sentence or group of sentences.
Without solid fundamentals of the basic sounds integral to a language, native-level fluency is impossible. Regardless of which level students are at, Speechling will critique audio at the granularity most applicable to the student in order to build a solid foundation.
The vision of Speechling is to create world class content for learning any language, from the spoken side.
We don't want to compete with language-learning giants on delivering tools for listening, reading, and writing.
However, our aspiration is to become the best language learning tool out there for being able to speak effectively, which currently remains the toughest and least explored of the four methods of communication.
The core tenets of the Speechling vision:
- Massive content collection. Our goal is to have a wide variety of sentences encompassing a wide array of domains from travel to medicine. Any-to-any language pair will be possible as the languages offered on the site expands.
- High quality feedback. Speechling will strive to offer pinpoint accuracy on the pronunciation feedback offered to students. This will be scaled to the content as it grows.
- Community engagement. We are available for you, no matter how many of you there are. We strive to build community help support through subreddits, Slack, and email to answer any questions and help building our product to its best potential.
We're excited for the launch of Speechling and the opportunities it provides to the language learning community.