REQUEST: Posting a supplementary audio clip

Question

(PLEASE UPVOTE IF YOU BELIEVE THE FEATURE REQUEST WOULD BENEFIT ELL)

Do you know what would be really cool? The possibility of recording our speech and posting it online, without leaving the ELL site. Why isn't this a thing?

The recording would be an optional supplement to the written answer. It needn't repeat word for word the written answer, it could be as simple as the pronunciation of a single word, or a short sentence in order to hear the difference pronunciation between an American and a British speaker. Questions about rhoticity and RP could be quickly and easily answered if we could upload an audio file, users could listen to the snippet without leaving the site, similar to the video clips posted on Movies&TV.SE.

Example: Is this bar slide trick shown on Cheers real or a visual effect?

But instead of visiting off-site (e.g. YouTube) and pasting a link, we could (by the powers of the Internet) use our own hand-held devices, laptops, tablets etc. to record ourselves and upload the audio clip, in a way similar to how we can now upload and an image.

Currently on ELL there 172 questions that have "How do I pronounce" in their titles, uploading an audio file in an answer would facilitate things enormously. However, no one should be obligated to record an audio file, it should be seen as a complementary asset, and in no way should it substitute a thoughtfully composed and well-supported answer.

Where is this technology?

Advantages

1. Hearing word stress and sentence intonation in different accents (British, American, Australian, Indian English etc.)
2. Hearing a correct (or several correct) pronunciations of difficult-to-pronounce English words.
3. Users wouldn't have to decipher IPA transcriptions, which to the uninitiated must appear to be gobbledegook (e.g. /ˈɡɒb(ə)ldɪˌɡuːk/)
4. Where there is no American, Australian, or British pronunciation of a particular word available on Forvo, or in an online dictionary, we could easily upload our own audio files and let people hear them without leaving the site.

If we are worried about new users abusing this capability, we could make this an earned privilege, users with a certain amount of rep would be trusted to use this feature sensibly. This feature request might even motivate users to continue posting high-quality posts in order to earn the privilege.

ADDENDUM

I have noted in some comments posted on ELL that some native users do not consider Indian English to be a valid variant of English. Some of these comments are derisive of its idiolect and argue that it is not real English.

Indian English is recognised world-wide as an English variant. A competent IndEn speaker should feel proud, never embarrassed, to record snippets of their speech. On Stack Exchange, inclusivity is encouraged, not dismissed.

UPDATE

Example: https://soundcloud.com/marilou-stu/my-voice

With this sample recording, you can "see" it is not necessary to read aloud every word, just the most relevant section. And perfection is not an absolute, e.g. I stumbled over the pronunciation of “uninitiated”. However, it would have been easy enough to record a new piece. All in all, it took me about 20 minutes from start to finish before I could upload the link. There is another drawback to this method, by clicking on the link the reader is forced to leave the site and can no longer read the answer while they are listening, to do so two tabs need to stay open if you are using a browser. I'm not sure how that would work on a smartphone as I'm not an expert on smartphones or devices.

Here is the updated version: https://soundcloud.com/marilou-stu/my-voice-2

One of the disadvantages mentioned says that it might be difficult to create high-quality recordings if one didn't have access to a quiet room and a decent microphone. Remember, an audio clip is not obligatory, and you wouldn't need specialized equipment to let everyone know how you pronounce a word like sixth or lunch. My audio snippet was recorded during the daytime on my smartphone, Samsung Galaxy 6, (so not the latest model) and I used a free app and no special microphone. The sound quality is good.

Tip: If recording, speak a little further away from the smartphone microphone than I did, that way users won't hear you breathing!

Answer 1

I think that adding audio recordings needs to be viewed similarly to adding an image. We discourage people from adding images to their answers just because it makes them more eye-catching, but sometimes they are really helpful. In my opinion, audio recordings need to have descriptions of their content, (just like images), not be too long (images shouldn't be too big), and not be used gratuitously (it doesn't make sense to record the full text of every answer - they should be supplements as explained in the question).

Audio is a little more difficult to moderate than imagery, but it's doable if we limit who can post audio to higher reputation users, have text descriptions of the content of the audio, and limit the length of the audio to about a sentence or two. There is technology to automatically attempt to transcribe audio, but I'm not sure how difficult it would be to incorporate and I don't think it's necessary.

Just to be clear - I assume audio clips will be restricted to answers from the way this discussion is phrased. I don't see them being as useful in questions.

Answer 2

1. The majority of language is spoken.

2. The priority, for a vast majority of English language learners, is to be able to converse in the spoken medium (involving both speaking and listening).

It seems perverse, therefore, to not let users use sound files in otherwise well constructed and well researched questions.

Of course, as with other questions that don't use sound files, we might get poor questions. But we can can get poor questions about anything and involving any medium. It is easy to sit around thinking about potential problems - anybody can do that. We need to think about what we are losing by not allowing sound files, and to my mind, we are losing rather a lot. This site, along with EL&U, is much poorer than it would otherwise be because we don't allow such files in posts.

I have used sound files in answers both here and on EL&U. Here, for example, is a question which it was not possible to answer convincingly for all readers without a sound file. However, there are several problems with this. Firstly, as you will find if you try to listen to the sound files, and as explained by Mari-Lou, you cannot hear the file without going to another page. Secondly, there is a very high probability of link-rot. The only trustworthy page I could find when writing that answer, does not keep sound files for very long, and I can't get the original phonetician to come and re-record it every time the original sound file disappears. In any other situation, we would exhort the poster to include a picture/sentence/quote/passage/example directly into their post. Here we find, perversely, posters being forced off site to convey their material to readers.

Some respondents here are fretting about sound quality and so forth. The sound files we made for the linked-to answer were made with a normal computer in a living room. All that was required was to turn the radio off. Many questions and answers here have had linked-to sound files, none of which, I am sure, were made using any specialist equipment whatsoever.

I urge you to up-vote Mari-Lou's question (and this answer, if you agree with its sentiment).

Answer 3

Please pinpoint the weaknesses and impracticable aspects of the FR

Before proposing this feature request to the Stack Exchange and its teams of developers the proposal needs to be watertight and we need to be ready to defend the idea to naysayers.

Needless to say, if the general response is lukewarm, the feature request won't progress beyond this stage.

Disadvantage #1

All the answers are protected as "sanctuary" because no edit is allowed except for answer, because if the written answer is changed by the 3rd party, the answer must make the sound changed too.

This observation echoes @whiskeychief's one, mainly that the Stack Exchange model encourages users, registered and none, to edit posts that need improvement. Whiskeychief and Kentaro Tomono suggest that every time a user, other than the author, edits a post it means the author has to re-record the answer.

Not necessarily so

If the correction consists of a typo (a simple spelling mistake) the recording remains valid. (see edit history in the original post)

The recording of a particular word is not affected if someone edits its spelling. For example, misspelling there as they're or their; the three words are homophones, which means they all have the same pronunciation. Fixing the spelling will not have any effect on the recording.

If the edit consists in correcting punctuation, capitalizing, formatting, using bold, italics, adding speech marks, or embedding a link, none of these edits will affect the recording.

In case this reminding, the recording can be very brief; mine was 50 seconds long but to supplement an answer on ELL one could record the stress pattern of a single sentence, which could be less than 10 seconds. Moreover, the user who has earned the privilege to add a recorded file will be someone who has at least 25K. In fact, Stack Exchange calls someone who has earned 25K a trusted user

Reputation also determines a user’s privileges within the system. As you gain more reputation, the system learns to trust you and bestows new functionality upon you that low-reputation users cannot access.

Which suggests that “trusted users” are regular contributors, people who consistently contribute to the community, and provide high-quality content that is well-supported. Although the edit may not have any adverse effect on the recording, authors are always notified if their post has been edited, in this way they can check whether the edit was helpful or not. If it was an inappropriate edit the author can roll the edit back to a previous revision.

Whenever an edit is suggested to a post you own, a notification with a link to both the post and the suggested edit will be sent your way. @Kevin Montrose♦ May 24 '12

disadvantage #2

J.R. noted the following aspect (in a deleted answer)

…audio recordings are difficult to edit, and the Stack Exchange is predicated on improvements made by the community. It's much easier to fix a typo in a written answer than fix a misspoken segment of an audio recording.

As mentioned earlier, fixing a typo will not affect the recording, and I agree editing an audio file is complicated but why edit when you could re-record the piece and instantly upload the file as you would with an image?

Unfortunately, it's unclear what is meant by “misspoken segment”. If it refers to a common mispronunciation, users can comment and later decide to downvote the answer. If “misspoken” refers to mumbling or misreading, it's reasonable to assume a trusted user would listen to their recording a couple of times before uploading it. I know I did, and I was aware that I fumbled over the word uninitiated, nothing prevented me from doing that recording a second time and then uploading the best version. Instead, I left the error, I did that recording in one go. No edits, no revisions, no fancy special effects. I did, however, take care not to speak too quickly, so learners could listen to how the word listen was emphasised, how words are pronounced and stressed, showing (hopefully) why English is a stressed-time language. You simply cannot do the same in a text answer.

Answer 4

I think this would be a bad idea for the following reasons:

Editing of questions and answers, which is fundamental to the success of the site, would make this unworkable. If the audio file was created by the OP and the question were to contain grammatical errors and require editing, the written question would no longer match the audio file. Questions and answers could go through multiple edits, and it seems unreasonable and an enormous amount of work to expect the OP (who in most cases is a non-native English speaker) or anybody else to continually re-record the audio in line with edits of the text.

Also, trying to comprehend a spoken question or answer with a variety of accents from non-native English speakers (the majority of OPs on this site), different pitch and pace, variable audio quality, and possibly an understandable number of mistakes from English learners, would be difficult and put many off answering. Native English speakers can read a lot faster and work through mistakes in text, even editing the question where this is helpful.

A user who reads a text answer and feels no need to listen to the audio file could upvote it based on the written content - yet the audio file could contain mispronunciations or other mistakes. For example, the sample audio file created by the OP and linked in their response to their own question has a clear mispronunciation of the word "dyslexic".

Hosting audio files would take up considerably more space than at present. Extremely low bitrates to save space would distort the audio, which would not be fit for purpose. If hosting the audio files became a future issue, countless questions would be lost.

If implemented for "trusted users" only (OP suggests reputation >25K) it would create a "class" divide that would unfairly disadvantage a new user who may be best placed to answer a particular question but not be able to provide audio; meanwhile anyone who has gained a reputation with a 'scattergun' approach and may not have the best answer will be able to give a multimedia answer which will appear "technically" superior.

Answer 5

Potential disadvantage 1

All the answers are protected as "sanctuary" because no edit is allowed except for answer, because if the written answer is changed by the 3rd party, the answer must make the sound changed too.

---

Potential disadvantage 2

There will be many questions with no accepted answer since every accent is equally "correct", which potentially could lessen the motivation of the answers.

---

Potential disadvantage 3

Now, the possible solution to the potential disadvantage 1 was presented to us. And here comes another problem. How would the OP balance the edit priviledge ( now rep 2K ) and the OP's proposed wall ( rep 25K ) and fill the gap?