I'm on the S0 Tier for Azure Cognitive Speech services and am trying to train a custom voice for Japanese TTS. My data was successfully processed. But I wasn't able to select "Statistical Parametric" or "Concatenative" as my training method. "Neural" was the only option on the list.
Training Method Options for Japanese model
However, I was able to use those non-neural methods for English and Chinese projects.
Training Method Options for English/Chinese model
Does anyone know if I could still train a Japanese non-neural voice model? If so, how?
Thank you very much in advance.
Non-Neutral voice training has been deprecated. The standard/non-neural training tier (adaptive, statistical parametric, concacenative) of Custom Voice is being deprecated. The announcement has been sent out to all existing Speech subscriptions before 2/28/2021. During the deprecation period (3/1/2021 - 2/29/2024), existing standard tier users can continue to use their non-neural models created. All new users/new speech resources should move to the neural tier/Custom Neural Voice. After 2/29/2024, all standard/non-neural custom voices will no longer be supported.
To answer your question about why you can train non-neutral for English and Chinese projects but not for Japanese: We allow current users to re-train non-neural voices for their existing projects, but they are not allowed to train new for non-neutral.
For user use non-neutral, we have a short guidance to help you mitigate to neutral training:
https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/how-to-custom-voice#migrate-to-custom-neural-voice
Sorry for the inconvenience.
Related
I'm seriously considering training a NMT model using some domain data. Custom Translator seems a good option.
But I'm wondering whether the model offered by Azure is an empty model (never trained with language data) or is pretrained general model (like Bing Translator)?
If it's the former case, do I need to train the model with additional general domain data to achieve ideal results.
Thanks
Custom Translator enables customer to use their own domain specific dataset to build translation engine that speaks their domain terminology. Customizing with general domain data would not deliver the translation quality the customer desires.
If customer does not have a domain dataset, it would be better to use our general domain (or standard models like bing translate). For more info about Custom Translator, https://techcommunity.microsoft.com/t5/azure-ai/customize-a-translation-to-make-sense-in-a-specific-context/ba-p/2811956
I see that there are advanced ML projects already that does text to speech such as SV2TTS : https://github.com/CorentinJ/Real-Time-Voice-Cloning
However what I am looking is rather than text to speech, re-synthesising speech into another voice
So are there any projects, software, library related to this subject that I can utilize?
I have found paid to have services but I need a free one for my experiments
Watson studio is a good software that includes libraries for speech recognition and speech translation. It might be able to help you with what you need to accomplish. I know that R has a built in library for this but it is very limited.
I got a big problem. For my bachelor thesis I have to make a machine tranlation model with BERT.
But I am not getting anywhere right now.
Do you know a documentation or something that can help me here?
I have read some papers in that direction but maybe there is a documentation or tutorial that can help me.
For my bachelor thesis I have to translate from a summary of a text into a title.
I hope someone can help me.
BERT is not a machine translation model, BERT is designed to provide a contextual sentence representation that should be useful for various NLP tasks. Although there exist ways how BERT can be incorporated into machine translation (https://openreview.net/forum?id=Hyl7ygStwB), it is not an easy problem and there are doubts if it really pays off.
From your question, it seems that you are not really machine translation, but automatic summarization. Similarly to machine translation, it can be approached using sequence-to-sequence models, but we do not call it translation in NLP.
For sequence-to-sequence modeling, there are different pre-trained models, such as BART or MASS. These should be much more useful than BERT.
Update in September 2022: There are multilingual BERT-like models, the most famous are multilingual BERT and XLM-RoBERTa. When fine-tuned carefully, they can be used as a universal encoder for machine translation and enable so-called zero-shot machine translation. The model is trained to translate from several source languages into English, but in the end, it can translate from all languages covered by the multilingual BERT-like models. The method is called SixT.
I'm trying to figure out the best approach to interfacing with Watson Conversation from a multi-language website (English/French). The input to Watson Conversation will be fed from Watson STT, so the input should be in the appropriate language. Should I set up the intents and entities in both languages? That might potentially cause issues with words that are the same (or very similar) in both languages but have different meanings. My guess is that I'll need two separate Conversation workspaces but that seems like a lot of overhead (double work when anything changes). I've thought about using the Watson Language Translator in between STT and Conversation but I would think the risk with that approach could be a reduction in accuracy. Has anyone been able to do this?
You will need to set up separate workspaces for each language. As you need to set the language of your workspace.
After that you would need to do STT, then language detection service to determine what workspace it should be directed to.
Which Microsoft Cognitive Services (or Azure Machine Learning services?) is best and least work to use to solve the problem of finding similar articles given an article. An article is a string of text. And assuming I do not have user interaction data about the articles.
Are there anything in Microsoft Cognitive Services that can solve this problem out-of-the-box? It seems I cannot use the Recommendations API since I don't have interaction/user data.
Anthony
I am not sure that Text Analytics API may be a good use for this scenario, at least not yet.
There are really two types of similarities:
1. Surface similarity (lexical) – Similarity by presence of words/alphabets
If we are looking for surface similarity, try fuzzy matching/lookup (SQL Server Integration Services – provides a component for this.), or approximate similarity functions (Jaro-Winkler distance, Levenshtein distance) etc. This would be easier as it would not require you to create a custom machine learning model.
2. Semantic similarity – Similarity by meaning of words
If we are looking for Semantic similarity, then you need to go for semantic clustering, word embedding, DSSM (Deep semantic similarity model) etc.
this is harder to do, as it would require you to train your own machine learning model based on an annotated corpus.
Luis Cabrera | Text Analytics Program Manager | Cloud AI Platform, Microsoft
Yes, you can use Text Analytics API.
examples are available here. https://www.microsoft.com/cognitive-services/en-us/text-analytics-api
I would suggest you use the Text Analytics API [1] as #Narasimha suggested. You would put your strings through the Topic detection API, and then come up with a metric (say, Similarity = count(Matching topics) - count(Non Matching topics)) that could order each string against the others for similarity. This would just require one API call and a little JSON parsing.
[1] https://www.microsoft.com/cognitive-services/en-us/text-analytics-api
Sentence similarity or semantic textual similarity is a measure of how similar two pieces of text are, or to what degree they express the same meaning.
This Microsoft's GitHub repo for NLP provide some sample wich could be used from Azure VM and Azure ML : https://github.com/microsoft/nlp/tree/master/examples/sentence_similarity
This folder contains examples and best practices, written in Jupyter notebooks, for building sentence similarity models. The gensen and pretrained embeddings utility scripts are used to speed up the model building process in the notebooks.
The sentence similarity scores can be used in a wide variety of applications, such as search/retrieval, nearest-neighbor or kernel-based classification methods, recommendations, and ranking tasks.