A new voice morphing application

Over the years, we've documented various applications of voice morphing technology besides the malicious creation of "deep fake" audio clips. Here's a new one: Amrit Dillon, "AI erases call centre staff’s Indian accents", The Times 3/2/2025:

A French company which operates the largest number of call centres in the world is using artificial intelligence to soften Indian accents in real time to make customer conversations easier and shorter.

Teleperformance said that it was sometimes difficult for customers calling call centres in India — and the Philippines — to understand workers’ accents, leading to frustration and longer than necessary calls.

“When you have an Indian agent on the line, sometimes it’s hard to hear, to understand,” Thomas Mackenbrock, the company’s deputy chief executive, told Bloomberg News. “The technology can neutralise the accent of the Indian speaker with zero latency. This creates more intimacy, increases customer satisfaction, and reduces the average handling time. It is a win-win for both parties.”

The software, called “accent translation”, has been developed by Sanas, a start-up based in Palo Alto, California.

Read the rest of this entry »

Comments (19)


Scream cipher

A recent xkcd:

Mouseover title: "AAAAAA A ÃA̧AȂA̦ ǍÅÂÃĀÁȂ AAAAAAA!"

Read the rest of this entry »

Comments (16)


"The Sway Mo' Blues"

[This is a guest post by Kirinputra.]

By now we know. The secret is out. The “Chinese character” is not as different as thought it was. We know sinographs don’t transcend the plane of sound — not quite like math symbols, anyway, or the man-woman bathroom icons, or stoplights.

How deep is the entanglement, though, between sinographs and sound? In contemplation of this, I present a fun-sized platter of puns and related matter from a seaborne sliver of the Sinosphere thought by some to be a living showcase of peak sinography: Formosa.

In particular, I want to spotlight a subconscious reading mechanism that guts the good of sinography and bends it to shady ends.

Let’s start with Sioumazang Yakiniku, Japanese barbeque chain. Sinographically, the name is 燒肉衆. The Mandarin reading, or name, is straightforward: Shāoròuzhòng. The Taioanese name is more or less undefined, despite the possibilities — for reasons that should be clear by the time we get to the other side of this.

Read the rest of this entry »

Comments (17)


Pronouncing DOGE

Coby L. wrote to ask why DOGE is pronounced with a final /ʒ/ rather than a final /dʒ/.

The Department Of Government Efficiency is clearly a backronym of the Doge meme, which references a Shiba Inu dog. According to Wikipedia, the meme can be pronounced /doʊʒ/ or /doʊdʒ/ or /doʊɡ/, though all I've heard from the media is /doʊʒ/. I guess Coby's experience is similar, hence the question. Wikipedia says that the memetic cryptocurrency Dogecoin is pronounced either /doʊʒkɔɪn/ or /doʊdʒkɔɪn/, but apparently not /doʊgkɔɪn/.

Read the rest of this entry »

Comments (21)


Another elephant in the room

Comments (18)


Pulmonic ingressive

Comments (12)


ChatGPT cat wound healing conversation

VHM:  This is a dialog held between ChatGPT and TK, who printed it out and sent it to me.

The unretouched dialog, which lasted about 20 minutes, is very long.  If you don't have time to read all of it, please look at the last paragraph of this post, where I give my takeaway assessment of the implications it holds for AI.

===================================================

TK:  It's truly amazing what ChatGPT does. I believe this is worth your time reading this conversation I had with it. Because not only is the topic and material interesting on its own, but also it will really give you a flavor and insight into how the AI known as ChatGPT functions.
 
Some people are claiming newer AIs are better but I can't see how this could be improved…. Only if the interface included sight and displayed images.
 
Note, I used voice dictation extensively in this conversation.
 
+++++++++++++++
 

Read the rest of this entry »

Comments (28)


AI Overview jokes

There's recently been a minor social- and mass-media fad for weird "AI Overview" answers from Google. The results are a moving target, either because of back-end fixes or because of the inherently stochastic nature of LLM results, but some of them are funny while they last. One query that still works this morning is a request for "elements that end in um but not ium", which sometimes answers

and sometimes answers


Read the rest of this entry »

Comments (21)


Fried and steamed mud: food for the season

From a Hong Kong restaurant:

Read the rest of this entry »

Comments (21)


Japanese "goods" | Chinese "guzi"

The pathways of word borrowings can be absolutely mind boggling.  The modern English word "goods" derives from the plural of one of six different roots that resulted in "good".  I will not touch upon the five other etyma that resulted in "good" with other meanings, but only on the one that culminated in a countable noun signifying "an item of merchandise", often fixed in the plural form "goods", e.g.,:

Inherited from Middle English good, god, from Old English gōd (a good thing, advantage, benefit, gift; good, goodness, welfare; virtue, ability, doughtiness; goods, property, wealth), from Proto-Germanic *gōdą (goods, belongings), from Proto-Indo-European *gʰedʰ-, *gʰodʰ- (to unite, be associated, suit). Compare German Gut (item of merchandise; estate; property).

(Wiktionary)

For two mercantile nations such as England and Japan, it is inevitable that "goods" would be borrowed from English into Japanese.  It has its own entry in the Japanese Wiktionary:  guzzu グッズ and has found its way into Korean as well:  gutjeu 굿즈.

Read the rest of this entry »

Comments (3)


Taiwan cares about its indigenous languages

AntC reports:

Earlier this month, I travelled the length of Taiwan's East Coast by train. Every station is decked out with Indigenous art, with the location's name given in both Mandarin and the applicable Indigenous language (Roman script).

Read the rest of this entry »

Comments (2)


Suzhou rap sounds like it has a French accent

From Chas Belov:  

Google Translate says that this song is in Suzhou topolect (it actually says "dialect" but thanks to you I know better). But I had to recognize a few words before I could convince myself it wasn't in French (which I also don't know). Later in the song it sounds more Chinese, but the rapper never really loses that French sound. Am I imagining things?

【苏州方言RAP】红中 Zyh 《三十三》PROD BY XVIBE

LISTEN HERE

Read the rest of this entry »

Comments (28)


ChatGPT does ASMR in Chinese

This morning, I received the following interesting message from Adam C.:

Back in 2019 you wrote a Language Log post about the word ASMR being ported to Japanese, and as I research the phenomenon itself I frequently encounter the same English version in videos by Japanese and Korean speakers. (Russians, unsurprisingly, use ACMP.)
 
So imagine my surprise at encountering the term 自主性感官經絡反應 on the Chinese ASMR Wikipedia page, which I understand is written in traditional characters. (I imagine the Taiwanese have taken over editing most of the Chinese Wikipedia because it's blocked on the mainland?)
 
Is there some sort of etymology for 自主性感官經絡反應, or is there anything else interesting about the phrase?

Read the rest of this entry »

Comments (1)