Get hands-on experience with 20+ free Google Cloud products and $300 in free credit for new customers.

Removed voices from German standard text to speech (tts)

We have a problem related to the Cloud text to speech API.
We develop an AI based chatbot system, and we have lot of different chatbot which speak in English and German also.
We are using two different voices 'de-DE-Standard-B' (male) and 'de-DE-Standard-C' (female) in the case of German bots, but both bots speak in same vois at now.
We detected the problem at 2022-02-16.

Could you give me some information about this problem?

0 3 479
3 REPLIES 3

Hello,

I understand you have selected 2 different German Language voices[ 'de-DE-Standard-B' (male) and 'de-DE-Standard-C' (female) ] from the list of available voices[0] that can be used for synthetic speech, however both voices are coming out the same.

As you have rightly indicated, these voices are different. However, it will be nice to understand how you are creating the voice audio files[1]. As indicated in the article[1], it is not only possible selecting a unique voice, you can also make certain modifications depending on your implementation. For example, you can modulate the output in pitch, volume, speaking rate, and sample rate. If you are using SSML in your audio synthesis, you would even have a finer-grain control over how the audio output.

So, please give more insight to your setup and how these voices are selected.

 

[0]https://cloud.google.com/text-to-speech/docs/voices

[1]https://cloud.google.com/text-to-speech/docs/create-audio#text-to-speech-ssml-java

Hi Oakinlaja - There seem to be an issue when using SSML to read date using the German langauge. It plays randomly different message. Please help. This is the below request to Google speech to text.

 

TTS Request JSON :: {"voice":{"ssmlGender":"MALE","name":"de-DE-Wavenet-E","languageCode":"de-DE"},"input":{"ssml":"<speak><say-as interpret-as=\"date\" format=\"yyyymmdd\"> 20220506<\/say-as><\/speak>"},"audioConfig":{"sampleRateHertz":8000,"volumeGainDb":0,"speakingRate":1,"audioEncoding":"LINEAR16","pitch":0,"effectsProfileId":["telephony-class-application"]}}

In English it works correctly. For english , i pass languageCode : En-US