Saturday, March 6, 2021

Xinhua Dictionary with English translation updated to latest version, hit market - Global Times - Translation

A child reads the new version of the Xinhua Zidian – Xinhua Dictionary (12th Edition) in the Beijing Book Building on Tuesday. The new edition was launched on Monday, adding new words such as chu xin” (mission), dian zan” (thumbs up) and “er wei ma” (QR code), making it the first time that an app and a paper book were issued simultaneously. The Xinhua Zidian is a Chinese language dictionary that was originally published in 1953. It is considered a symbol of Chinese culture. Photo: cnsphoto

 Photo: cnsphoto


China’s Commercial Press announced on Saturday the release of its latest version of the Chinese-English edition of the Xinhua Dictionary, which hit the market after major updates. 

As the most popular Chinese language dictionary, the new version includes more than 13,000 characters as well as 3,000 phrases, which proceed to explain words in contexts such as history, society and culture. 

The target readers of the dictionary are bilingual Chinese and English learners and translators. 

The handy language-learning tool has its main strengths such as being resourceful while simple to carry around, and the explanation of every single Chinese character corresponds to its English translation, with both of them being on the same page. 

Given the rich content, the dictionary is designed in an iconic red faux leather cover with golden gilded characters of ‘Xinhua Dictionary’ emblazoned in both Chinese and English

First published in 1953, the Chinese version of the Xinhua Dictionary is the most popular dictionary in China, which has had more than 600 million copies issued.

Global Times

Guide to Real-Time Face-to-Face Translation using LipSync GANs - Analytics India Magazine - Translation



State-of-the-art Neural Machine Translation systems have become increasingly competent in automatically translating natural languages. These systems have not only become formidable in plain text-to-text translation tasks but have also made a considerable leap in speech-to-speech translation tasks.  With the development of such systems, we are getting closer and closer to overcoming language barriers. However, there is still a medium that these systems need to tackle- videos.  As far as videos are concerned we are still stuck with transcripts, subtitles, and manual dubs. And the translation systems that do exist can only translate the audiovisual content at the speech-to-speech level. This creates two flaws- the translated voice sounds significantly different from the original speaker, the generated audio and the lip movements are unsynchronized. 

In their paper “Towards Automatic Face-to-Face Translation”, Prajwal K R et al tackle both these issues. They propose a new model LipGAN that generates realistic talking face videos across languages. And to work around the issue of personalizing the speaker’s voice, they make use of the CycleGAN architecture. 

Pipeline for Face-to-Face Translation

Pipeline for Face-to-Face Translation

In the very first phase of the pipeline, DeepSearch 2 Automatic Speech Recognition(ASR) model is used to transcribe the audio. To translate the text from language A to language B the Transformer-Base available in fairseq-py is re-implemented by training a multiway model to maximize learning. The trained model has parameters that are shared across seven languages – Hindi, English, Telugu, Malayalam, Tamil, Telugu, and Urdu.



DeepVoice 3 is employed for the text-to-speech(TTS) conversion, this model only generates the audio in one voice. A CycleGAN architecture model trained with 10 minutes of target’s audio clip is used to personalize the audio to match the voice of the target speaker.  

This personalized audio is passed to the lip-sync GAN, LipGAN, along with the frames from the original video.

LipGAN 

LipGAN architecture used in Face-to-Face Translation

The LipGAN generator network contains three branches-

  1. Face Encoder
    The encoder consists of residual blocks with intermediate down-sampling layers. Instead of passing a face image of a random pose and its corresponding audio segment to the generator, the LipGAN model inputs the target face with the bottom-half masked to act as a pose prior. This allows the generated face crops to be seamlessly pasted back into the original video without further post-processing.
  2. Audio Encoder
    LipGAN uses a standard CNN that takes a Mel-frequency cepstral coefficient (MFCC) heatmap for the audio encoder
  3. Face Decoder
    This branch takes the concatenated audio and face embeddings and creates a lip-synchronized face by inpainting the masked region of the input image with an appropriate mouth shape. It contains a series of residual blocks with a few intermediate deconvolutional layers that upsample the feature maps. The output layer of the decoder is a sigmoid activated 1×1 convolutional layer with 3 filters. 

The generator is trained to minimize L1 reconstruction loss between the generated frames and ground-truth frames

The discriminator network contains the same audio and face encoder as the generator network. It learns to detect synchronization by minimizing the following

contrastive loss:

Wav2Lip 

Since the “Towards Automatic Face-to-Face Translation” paper, the authors have come up with a better lip sync model Wav2Lip. The significant difference between the two is the discriminator. Wav2Lip uses a pre-trained lip-sync expert combined with a visual quality discriminator. 

See Also
LAMA - LightAutoML

The expert lip-sync discriminator is a modified, deeper SyncNet with residual connections trained on color images. It computes the dot product between the ReLU-activated video and speech embeddings. This yields the probability of the input audio-video pair being in sync:

Along with the L1 reconstruction loss, in Wav2Lip the generator is trained to also minimize the expert sync-loss

The visual quality discriminator consists of a stack of convolutional blocks. Each block consists of a convolutional layer followed by a leaky ReLU activation. It is trained to minimize the following objective function: 

Combining everything, the generator minimizes the weighted sum of the reconstruction(L1) loss, the synchronization loss (expert sync-loss), and the adversarial loss Lgen.

Speech to Lip Generation using Wav2Lip

  1. Install ffmpeg
    sudo apt-get install ffmpeg
  2. Create a new environment using either conda or venv
    conda create --name myenv  or python3 -m venv
  3. Clone the Wave2Lip repository
    git clone https://github.com/Rudrabha/Wav2Lip.git

  1. Move inside the Wave2Lip directory and install the necessary modules from the requirement.txt file
    cd Wav2Lip
    pip install -r requirements.txt

  2. Download the pre-trained GAN model from here and move it into the “Wav2Lip/checkpoints/”  folder
  3. Download the face detection model and put it in “face_detection/detection/sfd/” folder and rename it to “s3fd.pth

    wget "https://www.adrianbulat.com/downloads/python-fan/s3fd-619a316812.pth" -O "Wav2Lip/face_detection/detection/sfd/s3fd.pth"


  4. For the speech to lip generation to work it needs a video/image of the target face and a video/audio file containing the raw audio.

    python inference.py --checkpoint_path checkpoints/wav2lip_gan.pth --face "input.jpg" --audio "input.mp4"

    By default, the output video file named “result_voice.mp4” will be stored in the results folder, you can change this using the –outfile argument.

input image and audio/video for face-to-face translation
Output video of the face-to-face translation process

Subscribe to our Newsletter

Get the latest updates and relevant offers by sharing your email.
Join Our Telegram Group. Be part of an engaging online community. Join Here.

Tale of Immortal sells 1.8 million copies, devs “kicking off” English translation - PCGamesN - Translation

Tale of Immortal, the Chinese RPG game that’s been taking Steam by storm since it entered Early Access in January, has sold nearly two million copies in its first month. To celebrate, the team has started earnest work on an English translation, and wants your input.

In a post on Steam, developer Lightning Games says that 1.8 million people have bought the action-adventure game since launch. The team is “super happy to see how engaged our community is”, but knows there’s huge demand for a version in English, and so, the localisation process has now formally begun.

“From the next month, we will be kicking off the English localisation of the game!” reads the post. “For now, we would love to ask for your understanding before the translation is completed, we have set up a Discord Channel, and one of our team member will be answering any urgent questions/bug reports! We will also keep the translation progress updated in the channel.” The devs had said a translation was a priority back when it first started gaining popularity, but keeping the fantasy game stable amid hundreds of thousands of players likely took some precedence for the last several weeks.

You can join the official Discord here, if you’d like to contribute or discuss the localisation effort. “The team is currently working hard on fixing bugs, making the game more stable,” the post continues, “at the same time addressing issues some of our players have reported to maximise the game experience.”

Tale of Immortal has been consistently in Steam’s most played games, often ranking above GTA 5, Rust, and Rainbow Six Siege.

Some OxygenOS 11 users complain about broken translation and poor user experience - PiunikaWeb - Translation

Apparently, OnePlus has kept its focus limited to the looks of OxygenOS 11 and how well the UI feels with regards to single-handed usage. Hence, of course, they have done an excellent job there.

Gone are the days when OxygenOS was pretty much stock Android with just some ‘cherry-on-top’ feature additions designed to make stock Android richer than the barebones software it already is.

OxygenOS is now pretty unique in terms of overall feels with extremely modern-looking apps all following a simple principle – huge headers/action bars with large font sizes designed to push the top-most elements closer to the bottom to make single-handed operability easier.

oxygenos-11-one-handed

Some people will say that it now looks like a One UI rip-off but let’s not delve into that here. After all, if pulled off excellently, then why should anyone complain?

Now, everyone is aware of how the OEM is extremely keen on expanding its target market by diversifying its portfolio. Many of OnePlus’s recent offerings hint that the company has now drifted from its past philosophy of offering the best and nothing but the best.

With some competitive releases in the budget realms like the Nord N10 and N100, that too in select regions only, it is clear that OnePlus is now focusing on selling more in countries where they weren’t too popular historically.

And releases in more countries ‘translates’ to support for more languages in their software. OnePlus has done a good job with this regard, on paper at least. The OxygenOS 11 Wikipedia page shows that the OS supports over 40 different languages.

But just support for languages isn’t really sufficient. One has to get the translations right for a seamless and pleasant user experience. And this is where OnePlus has seemingly fallen short.

oxygenos-11-broken-translations-complaint

Source

Why russian-speakers have to deal with such poor user experience? We haven’t paid less for your smartphones. As a OnePlus 8 Pro user, for wich I paid about $1000, I’m very disappointed.
Source

There have been a bunch of recent complaints about broken Russian translation on OxygenOS 11. Users have shared several screenshots on the OnePlus forums showing incorrect grammar, hyphenation, and syntactical errors.

And the worse part is that these issues have plagued the entire interface and several stock apps as well so one doesn’t really have to be a grammar nazi or look too hard to spot them.

There are also plenty of screenshots revealing issues with fonts like incorrect or uneven text sizes or overflowing fonts. As already pointed out, OxygenOS 11 makes use of gigantic fonts, so there are plenty of instances of the latter.

oxygenos-11-broken-translations

Source

Users say that some of these issues have persisted for years. Many are also said to be due to incorrect strings in AOSP itself. But one can’t really blame AOSP for errors in OxygenOS apps.

And no, these issues are truly limited to languages other than the default one (English), since during our usage of OxygenOS 11 in the English language, we didn’t really find any errors as such.

All this doesn’t do well with OnePlus’s global image. Things like these may seem minor but are essential if OnePlus desires expansion.

Thus, the OEM should pull its socks up and fix the broken translation on OxygenOS 11 as soon as possible, before user dissatisfaction grows any further.

staff-member-comment

Source

Now, a OnePlus Staff Member did provide an explanation for it all. They even escalated some strings to the developers which proves that OnePlus is more than willing to fix such errors. Hopefully, we will eventually get to see OxygenOS 11 in the most perfect Russian with future updates.

For now, be sure to keep an eye on our OxygenOS 11 bug tracker so that you don’t miss out on anything.

PiunikaWeb started purely as an investigative tech journalism website with a main focus on ‘breaking’ or ‘exclusive’ news. In no time, our stories got picked up by the likes of Forbes, Fox News, Gizmodo, TechCrunch, Engadget, The Verge, MacRumors, and many others. Want to know more about us? Head here.

Friday, March 5, 2021

Between women, the dictionary and society - Sat, March 6 2021 - Jakarta Post - Dictionary

Posts on social media about the word perempuan (woman) drove me mad last month. The subentries of this word in the Great Dictionary of the Indonesian Language (KBBI) include perempuan geladak (prostitute), perempuan jalang (bitch), perempuan nakal (bad girl) and perempuan simpanan (mistress), which are all derogatory and may stigmatize women Many protesters have accused the dictionary’s drafting team of being blatantly misogynists. They assume the team has the authority to choose only positive phrases to be listed in the subentries. The Language Development and Fostering Agency, the composer and publisher of the dictionary, was held responsible. I almost joined the chorus of criticism, until I related it to my own experience in conducting linguistics research. As a researcher who gathers data from what is produced by speakers of a language in society, I have to make sure...

Women take fight against 'sexist' dictionaries to Italy - Thomson Reuters Foundation - Dictionary

Women's rights campaigner Maria Beatrice Giovanardi launches petition to change Italy's Treccani dictionary after successful push to amend Oxford Dictionaries' definition of 'woman'.

By Umberto Bacchi

MILAN, March 5 (Thomson Reuters Foundation) - Public figures from writers to lawmakers launched a campaign on Friday to change a leading Italian dictionary's "sexist" definition of a woman, which currently includes 30 different words for a sex worker.

About 100 high-profile Italians signed a letter demanding changes to the Treccani online dictionary after a similar campaign forced the Oxford English dictionary to alter its definition last year.

They argue that terms with negative connotations like "puttana" (whore) and "cagna" (bitch) should be dropped from a list of synonyms - and point out that the synonyms listed under "man" are broadly positive.

"Such expressions are not only offensive but ... reinforce negative and misogynist stereotypes that objectify women and present them as inferior beings," said the letter, published in the daily La Repubblica newspaper.

"This is dangerous as language shapes reality and influences the way women are perceived and treated."

Treccani did not immediately respond to a request for comment.

In an online post in November, the publisher said dictionaries recorded how words were used, and that any derogatory terms were labelled as such.

"The dictionary does not select lexicon based on moral judgment or prejudices," the post read. "If society and culture express negativity through words, a dictionary cannot refuse to document them."

In November last year, the Oxford University Press updated the definition of 'woman' in its dictionaries after a similar petition signed by tens of thousands of people sparked a review.

The renowned English language dictionary was criticised for listing terms such as "bitch", "bird" and "bint" as having a similar meaning to "woman".

Maria Beatrice Giovanardi, the equality activist who initiated both campaigns, said Treccani's definition was even more offensive, as it included 30 different terms to describe a sex worker.

"These words are simply not synonyms of the word 'woman'. They can be the offensive synonyms of the word 'sex worker', but not of 'woman'," she told the Thomson Reuters Foundation by phone.

"It's really a struggle to find anything positive in that definition, it's very outdated," added Giovanardi, an Italian national who lives in Britain.

Among the synonyms listed under the definition of man were "uomo d'affari" (businessman) "uomo di cuore" (man of heart) and "uomo d'ingegno" (man of genius), she said.

Giovanardi said she hoped the letter, signed by former lower house speaker Laura Boldrini and novelist Michela Murgia among others, would initiate a public debate on sexism in the Mediterranean country.

"Sexism is an everyday issue," she said "And dictionaries are first and foremost an educational tool."

Related stories:

'Our stories don't get told' – only now a sex workers’ podcast aims to change that 

Jamaicans share 'deepest secrets' in fresh push to allow abortion 

Lawsuits seen having 'chilling effect' on #MeToo movements in South Asia 

(Reporting by Umberto Bacchi @UmbertoBacchi, Editing by Claire Cozens. Please credit the Thomson Reuters Foundation, the charitable arm of Thomson Reuters, that covers the lives of people around the world who struggle to live freely or fairly. Visit http://news.trust.org)

Our Standards: The Thomson Reuters Trust Principles.

Syracuse grad tries to get ‘orbisculate,’ a word invented in CNY, added to dictionary - syracuse.com - Dictionary

A Syracuse University alumnus and his family are trying to get a word invented in Central New York added to the dictionary.

Jonathan Krieger, who graduated from SU’s Newhouse School in 2007, knows it’s not an easy task. But it’s an important one, as he seeks to honor his father, Neil Krieger, who died of complications from Covid-19 last year.

Neil Krieger created the word “orbisculate” for a class assignment while he was a student at Cornell University in Ithaca, N.Y., during the late 1950s. According to the official website orbisculate.com, it’s primarily a verb, with two definitions:

  • To accidentally squirt juice and/or pulp into one’s eye, as from a grapefruit when using a spoon to scoop out a section for eating. (Example: ”The grapefruit I was eating just orbisculated into my eye.”)
  • To accidentally squirt the inner content from fruits, vegetables and other foods onto one’s face, body or clothing, or onto that of a person nearby. (“I made a mistake dressing up before I ate a grapefruit. It ended up orbisculating on my shirt and now I have to change,” Jonathan Krieger told CNN as an example last month.)

There’s also a related noun, “orbisculation,” which describes the fruit juice itself: “Hey, you have something on your shirt.” “Oh no! That must be from the orbisculation of the orange I had earlier.”

Neil Krieger, who taught neuroscience at the University of Pennsylvania and Harvard medical schools and later started a biotech consulting company, loved the word and used it his whole life. His own children, Jonathan and Hilary, heard it so often they thought it was a real word.

Jonathan Krieger said he doesn’t know exactly when he learned it was a made-up word, but Hilary, a Cornell alum who works as an opinion editor at NBC News, remembers losing a $5 bet with a college friend over whether or not it was in the dictionary.

Neil Krieger had chronic kidney disease and was undergoing dialysis when he tested positive for the coronavirus in late March 2020. He was hospitalized for a month, dying of respiratory failure and complications due to Covid-19, according to medical records, on April 29. He was 78.

Jonathan Krieger and his sister launched their efforts to get “orbisculate” in the dictionary as a way of both grieving and celebrating their father.

“It’s just fun, it’s light, and that’s something that I think people could use right now, as opposed to something that gets a bit more serious,” he told CNN last month.

A petition seeks to add the word to English-language dictionaries in future editions. More than 5,700 people have signed as of Wednesday night, Krieger told syracuse.com.

Krieger, 35, lives in Brookline, Massachusetts, and runs the virtual events company Long Distance Trivia. He published a book about his varied career in 2018, titled “Odd Jobs: One Man’s Life Working Every Gig He Could Find, from Bathroom Attendant to Bikini Model.”

He and Hilary also created a list of 50 goals to popularize the word, which is the most important step in getting in the dictionary. (”Orbisculate” is already on UrbanDictionary.com, but Merriam-Webster and other major publications are harder to crack.)

Of those goals, they’ve already achieved nine, including getting the word in a crossword puzzle, engraved on a grapefruit spoon, and said in a podcast. They also hope to get it in a song (preferably by “Hamilton” star Lin-Manuel Miranda) or used by a celebrity with a fruit-y name, like “CNN anchor Don Lemon, fictional ’30 Rock’ character Liz Lemon, or Syracuse University mascot Otto the Orange.”

Jonathan Krieger is aware that Otto doesn’t talk, but the anthropomorphic orange is active on social media and could always tweet about orbisculate or hold up a sign with the word on it.

Krieger’s girlfriend, Megan O’Hara, also designed a logo and t-shirts featuring a cartoon citrus to help spread the word.

“She’s a really talented artist that can hopefully get more people on board,” Krieger said.

Sales from the shirts are all going to charity. More than $2,500 has been raised to benefit Carson’s Village, a Dallas non-profit that helps families after the loss of a loved one.

“...It feels fitting to honor our dad in a way that’s unique, that captures his humor and creativity and shares those attributes with the world,” Jonathan and Hilary wrote. “We suspect that the mission we’re embarking on may take years; if we’re being honest, we realize we may never accomplish our goal. But life has always been more about the journey than the destination. We know, because our dad taught us that.”

For more information or to sign the petition, visit orbisculate.com.