Learning patterns/A short guide to the recording of high-quality audio samples for Wiktionary/da
A short guide to recording of high-quality audio samples for Wiktionary
Recording audio samples for Wiktionary can present a considerable challenge to amateurs.
This learning patterns is supposed to present practical advice for recording audio samples for Wiktionary and encourage others to do so.
Please note: This learning pattern is based on texts of Wiktionary User:Jeuwre and was edited/ translated/ posted on Meta by Wikimedia Deutschland (WMDE).
Hvilket problem løser dette?
The right pronunciation is essential for non-native speakers, since the involvement of many different senses enables the learning process of new words, their meaning and usage in the brain. Therefore, audio samples are essential for any online dictionary. Unfortunately, Wiktionary still lacks many such audio samples. Users can only find such samples in other dictionaries.
Additionally, the recording of audio samples is not as easy as editing a Wiki page. It requires quite a bit more work in terms of preperation and editing. This learning pattern is supposed to offer practical advice and enable a quick introduction into the process. It is largely based upon texts of User:Jeuwre, who has recorded several thousand audio samples for the German Wiktionary so far.
Hvad er løsningen?
Please note: All pronunciation references used below refer to German language entries.
- A microphone of good quality is certainly essential. The built-in microphones in laptops or notebooks are inept for this task. Even normal headsets costing around 10 EUR or web cams with built-in microphones around 20 EUR are not suitable for recording because they produce too much white noise that can be annoying for the listener. Essential for the microphone is certainly the pop filter, which can limit the intial air push when people start speaking. This can be heard as a loud “popping” sound on a recording and can be quite annoying. German users, for example, can borrow such equipment from the Technikpool at Wikimedia Deutschland.
- A microphone stand or a ceiling mount can be very helpful (see Wiktionary:Pronunciation).
- If you do not have a mixer, you should pick a microphone with a USB connection, so that you can link it directly to your computer.
- Ideally, the microphone should have a a cardiod sensitivity pattern, so that only the voice of the speaker is being recorded and not its reflection of the walls and the ceilling.
- Possible candidates for a suitable microphones are, for example, a Rode NT-USB or a Blue Yeti. The first one has proven very useful in the recording project mentioned above.
- A long USB cable (5-6 meters) is very helpful in order to enable some distance between the microphone and the computer, since its cooling fan and hardware can produce noticeable noises on a recording.
- On your computer, you need a recording, editing and sound program. For this project, the free and under GPL licensed audio editing program Audacity proved very useful. Audacity is available for Linux, MS-Windows and Mac. It is easy to learn and very intuitive. You can find more information and the download link here.
General advice for making audio recordings
- If you don’t just want to record single words, you should make a list (on paper) of the desired phrases or sentences that you wish to record.
- Start small. A pragmatic way could be to pick a certain category and translate all terms from an area that you know well, for example Category:en:Musicians. Even though it can be quite frustrating to see how much is still left to record, completing all recordings for a certain category can be a big reward for your work.
- The pronunciation examples are supposed to reflect the natural language. So be aware not to overemphasize as in unnaturally enunciating and stretching syllables or strongly emphasizing certain letters.
- Record particular words a number of times if you think they don’t sound right the first time. For help, you can take a look at the International Phonetic Alphabet or listen to other recordings.
- Record while standing up. This makes your voice sound fuller and will cause less reflection from your computer screen or the walls. Put your notes with the list of words on a music stand, so that the paper does not rustle.
- The room has to be quiet. You will immediately hear ambient noise on your recording.
- Do not record anything that is not part of your active vocabulary. You should have a feeling for the word, its usage and context.
- Be conscious of your faulty pronunciation and language quirks. In German, a subconscious, but faulty pronunciation would be, for example, pronouncing a noun ending in -ng (such as the suffixes -ung and -ing) with a ‘k’ sound at the end: “Verwunderunk” (right: Verwunderung) or “Schmetterlink” (right: Schmetterling). People noticing such habits in their speech generally require higher concentration during recording sessions to avoid faulty pronunciations.
Additional tips for recording, editing, saving, uploading and embedding audio recording on particular sites
See the German help page on audio samples for further information.
Offline exchanges and group practice sessions
A Wiktionary meet-up with other active project editors and potentially interested speakers can be a great opportunity to practice and become motivated! You can exchange tips with other editors and often better evaluate online criticism on or suggestions for your work or particular articles. And, most importantly, you can find points of contact for your questions.
It can also be very helpful to practice and record audio samples in a group. This way, everyone can record the collectively selected list of words and receive direct feedback and advice regarding faulty pronunciations from the audience. For this, it can be helpful to project your editing program onto a big screen. A collective recording session can be fun without anyone being singled out in particular.
When to use
You should use this pattern if you would like to record audio samples for Wiktionary, but have never done a professional audio recording before.