Voice-specific prompting techniques - Amazon Nova

Voice-specific prompting techniques

Note

This documentation is for Amazon Nova Version 1. For the Amazon Nova 2 Speech-to-Speech prompt engineering guide, visit Voice conversation prompts.

Even within the same language, responses and word choices can be enhanced by steering the model for a particular locale. Therefore, we suggest that you include instructions in the system prompt to feature language that is natural and specific to the locale of your voice.

The available voices and locales are as follows:

Language

Feminine-sounding voice ID

Masculine-sounding voice ID

English (US)

tiffany

matthew

English (GB)

amy

French

ambre

florian

Italian

beatrice

lorenzo

German

greta

lennart

Spanish

lupe

carlos

British English

To specify British English usage, add the following line to your prompt:

Use British English as your language for your responses.

Example of a complete prompt with British English selection

You are a friend. The user and you will engage in a spoken dialog exchanging the transcripts of a natural real-time conversation. Keep your responses short, generally two or three sentences for chatty scenarios. Avoid formatted lists or numbering and keep your output as a spoken transcript to be acted out. Use British English as your language for your responses.

How to Specify a Voice Id

To select a specific voice for your interaction, include the voiceId parameter in your configuration. For example, to use the British English voice amy, use the following audioOutputConfiguration:

"audioOutputConfiguration": { "mediaType": "audio/lpcm", "sampleRateHertz": 24000, "sampleSizeBits": 16, "channelCount": 1, "voiceId": "amy", "encoding": "base64", "audioType": "SPEECH" }
Spanish

To specify Spanish as the response language, add the following line to your prompt:

Please respond exclusively in Spanish. If you have a question or suggestion, ask it in Spanish. I want to ensure that our communication remains in Spanish.

Example of a complete prompt with Spanish selection

You are a friend. The user and you will engage in a spoken dialog exchanging the transcripts of a natural real-time conversation. Keep your responses short, generally two or three sentences for chatty scenarios. Avoid formatted lists or numbering and keep your output as a spoken transcript to be acted out. Please respond exclusively in Spanish. If you have a question or suggestion, ask it in Spanish. I want to ensure that our communication remains in Spanish.

How to Specify a Voice Id

To select a specific voice for your interaction, include the voiceId parameter in your configuration. For example, to use the Spanish voice carlos, use the following audioOutputConfiguration:

"audioOutputConfiguration": { "mediaType": "audio/lpcm", "sampleRateHertz": 24000, "sampleSizeBits": 16, "channelCount": 1, "voiceId": "carlos", "encoding": "base64", "audioType": "SPEECH" }
French

To specify French as the response language, add the following line to your prompt:

Please respond exclusively in French. If you have a question or suggestion, ask it in French. I want to ensure that our communication remains in French.

Example of a complete prompt with French selection

You are a friend. The user and you will engage in a spoken dialog exchanging the transcripts of a natural real-time conversation. Keep your responses short, generally two or three sentences for chatty scenarios. Avoid formatted lists or numbering and keep your output as a spoken transcript to be acted out. Please respond exclusively in French. If you have a question or suggestion, ask it in French. I want to ensure that our communication remains in French.

How to Specify a Voice Id

To select a specific voice for your interaction, include the voiceId parameter in your configuration. For example, to use the French voice ambre, use the following audioOutputConfiguration:

"audioOutputConfiguration": { "mediaType": "audio/lpcm", "sampleRateHertz": 24000, "sampleSizeBits": 16, "channelCount": 1, "voiceId": "ambre", "encoding": "base64", "audioType": "SPEECH" }
Italian

To specify Italian as the response language, add the following line to your prompt:

Please respond exclusively in Italian. If you have a question or suggestion, ask it in Italian. I want to ensure that our communication remains in Italian.

Example of a complete prompt with Italian selection

You are a friend. The user and you will engage in a spoken dialog exchanging the transcripts of a natural real-time conversation. Keep your responses short, generally two or three sentences for chatty scenarios. Avoid formatted lists or numbering and keep your output as a spoken transcript to be acted out. Please respond exclusively in Italian. If you have a question or suggestion, ask it in Italian. I want to ensure that our communication remains in Italian.

How to Specify a Voice Id

To select a specific voice for your interaction, include the voiceId parameter in your configuration. For example, to use the Italian voice lorenzo, use the following audioOutputConfiguration:

"audioOutputConfiguration": { "mediaType": "audio/lpcm", "sampleRateHertz": 24000, "sampleSizeBits": 16, "channelCount": 1, "voiceId": "lorenzo", "encoding": "base64", "audioType": "SPEECH" }
German

To specify German as the response language, add the following line to your prompt:

Please respond exclusively in German. If you have a question or suggestion, ask it in German. I want to ensure that our communication remains in German.

Example of a complete prompt with German selection

You are a friend. The user and you will engage in a spoken dialog exchanging the transcripts of a natural real-time conversation. Keep your responses short, generally two or three sentences for chatty scenarios. Avoid formatted lists or numbering and keep your output as a spoken transcript to be acted out. Please respond exclusively in German. If you have a question or suggestion, ask it in German. I want to ensure that our communication remains in German.

How to Specify a Voice Id

To select a specific voice for your interaction, include the voiceId parameter in your configuration. For example, to use the German voice greta, use the following audioOutputConfiguration:

"audioOutputConfiguration": { "mediaType": "audio/lpcm", "sampleRateHertz": 24000, "sampleSizeBits": 16, "channelCount": 1, "voiceId": "greta", "encoding": "base64", "audioType": "SPEECH" }