Gemini-TTS

  import 
  
 os 
 from 
  
 google.cloud 
  
 import 
 texttospeech 
 PROJECT_ID 
 = 
 os 
 . 
 getenv 
 ( 
 "GOOGLE_CLOUD_PROJECT" 
 ) 
 TTS_LOCATION 
 = 
 os 
 . 
 getenv 
 ( 
 "GOOGLE_CLOUD_REGION" 
 ) 
 API_ENDPOINT 
 = 
 ( 
 f 
 " 
 { 
 TTS_LOCATION 
 } 
 -texttospeech.googleapis.com" 
 if 
 TTS_LOCATION 
 != 
 "global" 
 else 
 "texttospeech.googleapis.com" 
 ) 
 client 
 = 
 texttospeech 
 . 
  TextToSpeechClient 
 
 ( 
 client_options 
 = 
 ClientOptions 
 ( 
 api_endpoint 
 = 
 API_ENDPOINT 
 ) 
 ) 
 

  GLOBAL_API_ENDPOINT 
 = 
https://texttospeech.googleapis.com REGIONAL_API_ENDPOINT 
 = 
https:// YOUR_REGION 
-texttospeech.googleapis.com 

  # google-cloud-texttospeech minimum version 2.29.0 is required. 
 import 
  
 os 
 from 
  
 google.cloud 
  
 import 
 texttospeech 
 PROJECT_ID 
 = 
 os 
 . 
 getenv 
 ( 
 "GOOGLE_CLOUD_PROJECT" 
 ) 
 def 
  
 synthesize 
 ( 
 prompt 
 : 
 str 
 , 
 text 
 : 
 str 
 , 
 output_filepath 
 : 
 str 
 = 
 "output.mp3" 
 ): 
  
 """Synthesizes speech from the input text and saves it to an MP3 file. 
 Args: 
 prompt: Styling instructions on how to synthesize the content in 
 the text field. 
 text: The text to synthesize. 
 output_filepath: The path to save the generated audio file. 
 Defaults to "output.mp3". 
 """ 
 client 
 = 
 texttospeech 
 . 
  TextToSpeechClient 
 
 () 
 synthesis_input 
 = 
 texttospeech 
 . 
  SynthesisInput 
 
 ( 
 text 
 = 
 text 
 , 
 prompt 
 = 
 prompt 
 ) 
 # Select the voice you want to use. 
 voice 
 = 
 texttospeech 
 . 
  VoiceSelectionParams 
 
 ( 
 language_code 
 = 
 "en-US" 
 , 
 name 
 = 
 "Charon" 
 , 
 # Example voice, adjust as needed 
 model_name 
 = 
 "gemini-2.5-pro-tts" 
 ) 
 audio_config 
 = 
 texttospeech 
 . 
  AudioConfig 
 
 ( 
 audio_encoding 
 = 
 texttospeech 
 . 
  AudioEncoding 
 
 . 
 MP3 
 ) 
 # Perform the text-to-speech request on the text input with the selected 
 # voice parameters and audio file type. 
 response 
 = 
 client 
 . 
  synthesize_speech 
 
 ( 
 input 
 = 
 synthesis_input 
 , 
 voice 
 = 
 voice 
 , 
 audio_config 
 = 
 audio_config 
 ) 
 # The response's audio_content is binary. 
 with 
 open 
 ( 
 output_filepath 
 , 
 "wb" 
 ) 
 as 
 out 
 : 
 out 
 . 
 write 
 ( 
 response 
 . 
 audio_content 
 ) 
 print 
 ( 
 f 
 "Audio content written to file: 
 { 
 output_filepath 
 } 
 " 
 ) 
 

  # Make sure to install gcloud cli, and sign in to your project. 
 # Make sure to use your PROJECT_ID value. 
 # The available models are gemini-2.5-flash-tts, gemini-2.5-flash-lite-preview-tts, and gemini-2.5-pro-tts. 
 # To parse the JSON output and use it directly see the last line of the command. 
 # Requires JQ and ffplay library to be installed. 
 PROJECT_ID 
 = 
 YOUR_PROJECT_ID 
curl  
-X  
POST  
 \ 
  
-H  
 "Authorization: Bearer 
 $( 
gcloud  
auth  
application-default  
print-access-token ) 
 " 
  
 \ 
  
-H  
 "x-goog-user-project: 
 $PROJECT_ID 
 " 
  
 \ 
  
-H  
 "Content-Type: application/json" 
  
 \ 
-d  
 '{ 
 "input": { 
 "prompt": "Say the following in a curious way", 
 "text": "OK, so... tell me about this [uhm] AI thing." 
 }, 
 "voice": { 
 "languageCode": "en-us", 
 "name": "Kore", 
 "model_name": "gemini-2.5-flash-tts" 
 }, 
 "audioConfig": { 
 "audioEncoding": "LINEAR16" 
 } 
 }' 
  
 \ 
  
 "https://texttospeech.googleapis.com/v1/text:synthesize" 
  
 \ 
  
 | 
  
jq  
-r  
 '.audioContent' 
  
 | 
  
base64  
-d  
 | 
  
ffplay  
-  
-autoexit 

  # google-cloud-texttospeech minimum version 2.29.0 is required. 
 import 
  
 datetime 
 import 
  
 os 
 import 
  
 numpy 
  
 as 
  
 np 
 from 
  
 google.cloud 
  
 import 
 texttospeech 
 PROJECT_ID 
 = 
 os 
 . 
 getenv 
 ( 
 "GOOGLE_CLOUD_PROJECT" 
 ) 
 def 
  
 synthesize 
 ( 
 prompt 
 : 
 str 
 , 
 text_chunks 
 : 
 list 
 [ 
 str 
 ], 
 model 
 : 
 str 
 , 
 voice 
 : 
 str 
 , 
 locale 
 : 
 str 
 ): 
  
 """Synthesizes speech from the input text. 
 Args: 
 prompt: Styling instructions on how to synthesize the content in 
 the text field. 
 text_chunks: Text chunks to synthesize. Note that The synthesis will 
 start when the client initiates half-close. 
 model: gemini tts model name. gemini-2.5-flash-tts, gemini-2.5-flash-lite-preview-tts, and gemini-2.5-pro-tts 
 voice: voice name. Example: leda, kore. Refer to available voices 
 locale: locale name. Example: en-us. Refer to available locales. 
 """ 
 client 
 = 
 texttospeech 
 . 
  TextToSpeechClient 
 
 () 
 config_request 
 = 
 texttospeech 
 . 
  StreamingSynthesizeRequest 
 
 ( 
 streaming_config 
 = 
 texttospeech 
 . 
  StreamingSynthesizeConfig 
 
 ( 
 voice 
 = 
 texttospeech 
 . 
  VoiceSelectionParams 
 
 ( 
 name 
 = 
 voice 
 , 
 language_code 
 = 
 locale 
 , 
 model_name 
 = 
 model 
 ) 
 ) 
 ) 
 # Example request generator. A function like this can be linked to an LLM 
 # text generator and the text can be passed to the TTS API asynchronously. 
 def 
  
 request_generator 
 (): 
 yield 
 config_request 
 for 
 i 
 , 
 text 
 in 
 enumerate 
 ( 
 text_chunks 
 ): 
 yield 
 texttospeech 
 . 
  StreamingSynthesizeRequest 
 
 ( 
 input 
 = 
 texttospeech 
 . 
  StreamingSynthesisInput 
 
 ( 
 text 
 = 
 text 
 , 
 # Prompt is only supported in the first input chunk. 
 prompt 
 = 
 prompt 
 if 
 i 
 == 
 0 
 else 
 None 
 , 
 ) 
 ) 
 request_start_time 
 = 
 datetime 
 . 
 datetime 
 . 
 now 
 () 
 streaming_responses 
 = 
 client 
 . 
  streaming_synthesize 
 
 ( 
 request_generator 
 ()) 
 is_first_chunk_received 
 = 
 False 
 final_audio_data 
 = 
 np 
 . 
 array 
 ([]) 
 num_chunks_received 
 = 
 0 
 for 
 response 
 in 
 streaming_responses 
 : 
 # just a simple progress indicator 
 num_chunks_received 
 += 
 1 
 print 
 ( 
 "." 
 , 
 end 
 = 
 "" 
 ) 
 if 
 num_chunks_received 
 % 
 40 
 == 
 0 
 : 
 print 
 ( 
 "" 
 ) 
 # measuring time to first audio 
 if 
 not 
 is_first_chunk_received 
 : 
 is_first_chunk_received 
 = 
 True 
 first_chunk_received_time 
 = 
 datetime 
 . 
 datetime 
 . 
 now 
 () 
 # accumulating audio. In a web-server scenario, you would want to 
 # "emit" audio to the frontend as soon as it arrives. 
 # 
 # For example using flask socketio, you could do the following 
 # from flask_socketio import SocketIO, emit 
 # emit("audio", response.audio_content) 
 # socketio.sleep(0) 
 audio_data 
 = 
 np 
 . 
 frombuffer 
 ( 
 response 
 . 
 audio_content 
 , 
 dtype 
 = 
 np 
 . 
 int16 
 ) 
 final_audio_data 
 = 
 np 
 . 
 concatenate 
 (( 
 final_audio_data 
 , 
 audio_data 
 )) 
 time_to_first_audio 
 = 
 first_chunk_received_time 
 - 
 request_start_time 
 time_to_completion 
 = 
 datetime 
 . 
 datetime 
 . 
 now 
 () 
 - 
 request_start_time 
 audio_duration 
 = 
 len 
 ( 
 final_audio_data 
 ) 
 / 
 24_000 
 # default sampling rate. 
 print 
 ( 
 " 
 \n 
 " 
 ) 
 print 
 ( 
 f 
 "Time to first audio: 
 { 
 time_to_first_audio 
 . 
 total_seconds 
 () 
 } 
 seconds" 
 ) 
 print 
 ( 
 f 
 "Time to completion: 
 { 
 time_to_completion 
 . 
 total_seconds 
 () 
 } 
 seconds" 
 ) 
 print 
 ( 
 f 
 "Audio duration: 
 { 
 audio_duration 
 } 
 seconds" 
 ) 
 return 
 final_audio_data 
 

  # google-cloud-texttospeech minimum version 2.31.0 is required. 
 import 
  
 os 
 from 
  
 google.cloud 
  
 import 
 texttospeech 
 PROJECT_ID 
 = 
 os 
 . 
 getenv 
 ( 
 "GOOGLE_CLOUD_PROJECT" 
 ) 
 def 
  
 synthesize_multispeaker_freeform 
 ( 
 prompt 
 : 
 str 
 , 
 text 
 : 
 str 
 , 
 output_filepath 
 : 
 str 
 = 
 "output_non_turn_based.wav" 
 , 
 ): 
  
 """Synthesizes speech from non-turn-based input and saves it to a WAV file. 
 Args: 
 prompt: Styling instructions on how to synthesize the content in the 
 text field. 
 text: The text to synthesize, containing speaker aliases to indicate 
 different speakers. Example: "Sam: Hi Bob!\nBob: Hi Sam!" 
 output_filepath: The path to save the generated audio file. Defaults to 
 "output_non_turn_based.wav". 
 """ 
 client 
 = 
 texttospeech 
 . 
  TextToSpeechClient 
 
 () 
 synthesis_input 
 = 
 texttospeech 
 . 
  SynthesisInput 
 
 ( 
 text 
 = 
 text 
 , 
 prompt 
 = 
 prompt 
 ) 
 multi_speaker_voice_config 
 = 
 texttospeech 
 . 
  MultiSpeakerVoiceConfig 
 
 ( 
 speaker_voice_configs 
 = 
 [ 
 texttospeech 
 . 
  MultispeakerPrebuiltVoice 
 
 ( 
 speaker_alias 
 = 
 "Speaker1" 
 , 
 speaker_id 
 = 
 "Kore" 
 , 
 ), 
 texttospeech 
 . 
  MultispeakerPrebuiltVoice 
 
 ( 
 speaker_alias 
 = 
 "Speaker2" 
 , 
 speaker_id 
 = 
 "Charon" 
 , 
 ), 
 ] 
 ) 
 voice 
 = 
 texttospeech 
 . 
  VoiceSelectionParams 
 
 ( 
 language_code 
 = 
 "en-US" 
 , 
 model_name 
 = 
 "gemini-2.5-pro-tts" 
 , 
 multi_speaker_voice_config 
 = 
 multi_speaker_voice_config 
 , 
 ) 
 audio_config 
 = 
 texttospeech 
 . 
  AudioConfig 
 
 ( 
 audio_encoding 
 = 
 texttospeech 
 . 
  AudioEncoding 
 
 . 
 LINEAR16 
 , 
 sample_rate_hertz 
 = 
 24000 
 , 
 ) 
 response 
 = 
 client 
 . 
  synthesize_speech 
 
 ( 
 input 
 = 
 synthesis_input 
 , 
 voice 
 = 
 voice 
 , 
 audio_config 
 = 
 audio_config 
 ) 
 with 
 open 
 ( 
 output_filepath 
 , 
 "wb" 
 ) 
 as 
 out 
 : 
 out 
 . 
 write 
 ( 
 response 
 . 
 audio_content 
 ) 
 print 
 ( 
 f 
 "Audio content written to file: 
 { 
 output_filepath 
 } 
 " 
 ) 
 

  # Make sure to install gcloud cli, and sign in to your project. 
 # Make sure to use your PROJECT_ID value. 
 # The available models are gemini-2.5-flash-tts and gemini-2.5-pro-tts 
 # To parse the JSON output and use it directly see the last line of the command. 
 # Requires JQ and ffplay library to be installed. 
 # google-cloud-texttospeech minimum version 2.31.0 is required. 
 PROJECT_ID 
 = 
 YOUR_PROJECT_ID 
curl  
-X  
POST  
 \ 
  
-H  
 "Authorization: Bearer 
 $( 
gcloud  
auth  
application-default  
print-access-token ) 
 " 
  
 \ 
  
-H  
 "x-goog-user-project: 
 $PROJECT_ID 
 " 
  
 \ 
  
-H  
 "Content-Type: application/json" 
  
 \ 
-d  
 '{ 
 "input": { 
 "prompt": "Say the following as a conversation between friends.", 
 "text": "Sam: Hi Bob, how are you?\\nBob: I am doing well, and you?" 
 }, 
 "voice": { 
 "languageCode": "en-us", 
 "modelName": "gemini-2.5-flash-tts", 
 "multiSpeakerVoiceConfig": { 
 "speakerVoiceConfigs": [ 
 { 
 "speakerAlias": "Sam", 
 "speakerId": "Kore" 
 }, 
 { 
 "speakerAlias": "Bob", 
 "speakerId": "Charon" 
 } 
 ] 
 } 
 }, 
 "audioConfig": { 
 "audioEncoding": "LINEAR16", 
 "sampleRateHertz": 24000 
 } 
 }' 
  
 \ 
  
 "https://texttospeech.googleapis.com/v1/text:synthesize" 
  
 \ 
  
 | 
  
jq  
-r  
 '.audioContent' 
  
 | 
  
base64  
-d  
 | 
  
ffplay  
-  
-autoexit 

  # google-cloud-texttospeech minimum version 2.31.0 is required. 
 import 
  
 os 
 from 
  
 google.cloud 
  
 import 
 texttospeech 
 PROJECT_ID 
 = 
 os 
 . 
 getenv 
 ( 
 "GOOGLE_CLOUD_PROJECT" 
 ) 
 def 
  
 synthesize_multispeaker_structured 
 ( 
 prompt 
 : 
 str 
 , 
 turns 
 : 
 list 
 [ 
 texttospeech 
 . 
  MultiSpeakerMarkup 
 
 . 
  Turn 
 
 ], 
 output_filepath 
 : 
 str 
 = 
 "output_turn_based.wav" 
 , 
 ): 
  
 """Synthesizes speech from turn-based input and saves it to a WAV file. 
 Args: 
 prompt: Styling instructions on how to synthesize the content in the 
 text field. 
 turns: A list of texttospeech.MultiSpeakerMarkup.Turn objects representing 
 the dialogue turns. 
 output_filepath: The path to save the generated audio file. Defaults to 
 "output_turn_based.wav". 
 """ 
 client 
 = 
 texttospeech 
 . 
  TextToSpeechClient 
 
 () 
 synthesis_input 
 = 
 texttospeech 
 . 
  SynthesisInput 
 
 ( 
 multi_speaker_markup 
 = 
 texttospeech 
 . 
  MultiSpeakerMarkup 
 
 ( 
 turns 
 = 
 turns 
 ), 
 prompt 
 = 
 prompt 
 , 
 ) 
 multi_speaker_voice_config 
 = 
 texttospeech 
 . 
  MultiSpeakerVoiceConfig 
 
 ( 
 speaker_voice_configs 
 = 
 [ 
 texttospeech 
 . 
  MultispeakerPrebuiltVoice 
 
 ( 
 speaker_alias 
 = 
 "Speaker1" 
 , 
 speaker_id 
 = 
 "Kore" 
 , 
 ), 
 texttospeech 
 . 
  MultispeakerPrebuiltVoice 
 
 ( 
 speaker_alias 
 = 
 "Speaker2" 
 , 
 speaker_id 
 = 
 "Charon" 
 , 
 ), 
 ] 
 ) 
 voice 
 = 
 texttospeech 
 . 
  VoiceSelectionParams 
 
 ( 
 language_code 
 = 
 "en-US" 
 , 
 model_name 
 = 
 "gemini-2.5-pro-tts" 
 , 
 multi_speaker_voice_config 
 = 
 multi_speaker_voice_config 
 , 
 ) 
 audio_config 
 = 
 texttospeech 
 . 
  AudioConfig 
 
 ( 
 audio_encoding 
 = 
 texttospeech 
 . 
  AudioEncoding 
 
 . 
 LINEAR16 
 , 
 sample_rate_hertz 
 = 
 24000 
 , 
 ) 
 response 
 = 
 client 
 . 
  synthesize_speech 
 
 ( 
 input 
 = 
 synthesis_input 
 , 
 voice 
 = 
 voice 
 , 
 audio_config 
 = 
 audio_config 
 ) 
 with 
 open 
 ( 
 output_filepath 
 , 
 "wb" 
 ) 
 as 
 out 
 : 
 out 
 . 
 write 
 ( 
 response 
 . 
 audio_content 
 ) 
 print 
 ( 
 f 
 "Audio content written to file: 
 { 
 output_filepath 
 } 
 " 
 ) 
 

  # Make sure to install gcloud cli, and sign in to your project. 
 # Make sure to use your PROJECT_ID value. 
 # The available models are gemini-2.5-flash-tts and gemini-2.5-pro-tts. 
 # To parse the JSON output and use it directly see the last line of the command. 
 # Requires JQ and ffplay library to be installed. 
 # google-cloud-texttospeech minimum version 2.31.0 is required. 
 PROJECT_ID 
 = 
 YOUR_PROJECT_ID 
curl  
-X  
POST  
 \ 
  
-H  
 "Authorization: Bearer 
 $( 
gcloud  
auth  
application-default  
print-access-token ) 
 " 
  
 \ 
  
-H  
 "x-goog-user-project: 
 $PROJECT_ID 
 " 
  
 \ 
  
-H  
 "Content-Type: application/json" 
  
 \ 
-d  
 '{ 
 "input": { 
 "prompt": "Say the following as a conversation between friends.", 
 "multiSpeakerMarkup": { 
 "turns": [ 
 { 
 "speaker": "Sam", 
 "text": "Hi Bob, how are you?" 
 }, 
 { 
 "speaker": "Bob", 
 "text": "I am doing well, and you?" 
 } 
 ] 
 } 
 }, 
 "voice": { 
 "languageCode": "en-us", 
 "modelName": "gemini-2.5-flash-tts", 
 "multiSpeakerVoiceConfig": { 
 "speakerVoiceConfigs": [ 
 { 
 "speakerAlias": "Sam", 
 "speakerId": "Kore" 
 }, 
 { 
 "speakerAlias": "Bob", 
 "speakerId": "Charon" 
 } 
 ] 
 } 
 }, 
 "audioConfig": { 
 "audioEncoding": "LINEAR16", 
 "sampleRateHertz": 24000 
 } 
 }' 
  
 \ 
  
 "https://texttospeech.googleapis.com/v1/text:synthesize" 
  
 \ 
  
 | 
  
jq  
-r  
 '.audioContent' 
  
 | 
  
base64  
-d  
 | 
  
ffplay  
-  
-autoexit 

  from 
  
 google 
  
 import 
 genai 
 from 
  
 google.genai 
  
 import 
 types 
 import 
  
 wave 
 import 
  
 os 
 PROJECT_ID 
 = 
 os 
 . 
 getenv 
 ( 
 "GOOGLE_CLOUD_PROJECT" 
 ) 
 LOCATION 
 = 
 os 
 . 
 environ 
 . 
 get 
 ( 
 "GOOGLE_CLOUD_REGION" 
 , 
 "global" 
 ) 
 # Set up the wave file to save the output: 
 def 
  
 wave_file 
 ( 
 filename 
 , 
 pcm 
 , 
 channels 
 = 
 1 
 , 
 rate 
 = 
 24000 
 , 
 sample_width 
 = 
 2 
 ): 
 with 
 wave 
 . 
 open 
 ( 
 filename 
 , 
 "wb" 
 ) 
 as 
 wf 
 : 
 wf 
 . 
 setnchannels 
 ( 
 channels 
 ) 
 wf 
 . 
 setsampwidth 
 ( 
 sample_width 
 ) 
 wf 
 . 
 setframerate 
 ( 
 rate 
 ) 
 wf 
 . 
 writeframes 
 ( 
 pcm 
 ) 
 client 
 = 
 genai 
 . 
 Client 
 ( 
 vertexai 
 = 
 True 
 , 
 project 
 = 
 PROJECT_ID 
 , 
 location 
 = 
 LOCATION 
 ) 
 response 
 = 
 client 
 . 
 models 
 . 
 generate_content 
 ( 
 model 
 = 
 "gemini-2.5-flash-tts" 
 , 
 contents 
 = 
 "Say the following in a curious way: OK, so... tell me about this [uhm] AI thing." 
 , 
 config 
 = 
 types 
 . 
 GenerateContentConfig 
 ( 
 speech_config 
 = 
 types 
 . 
 SpeechConfig 
 ( 
 language_code 
 = 
 "en-in" 
 , 
 voice_config 
 = 
 types 
 . 
 VoiceConfig 
 ( 
 prebuilt_voice_config 
 = 
 types 
 . 
 PrebuiltVoiceConfig 
 ( 
 voice_name 
 = 
 'Kore' 
 , 
 ) 
 ) 
 ), 
 ) 
 ) 
 data 
 = 
 response 
 . 
 candidates 
 [ 
 0 
 ] 
 . 
 content 
 . 
 parts 
 [ 
 0 
 ] 
 . 
 inline_data 
 . 
 data 
 file_name 
 = 
 'output_speech.wav' 
 wave_file 
 ( 
 file_name 
 , 
 data 
 ) 
 # Saves the file to current directory 
 

  # Make sure to install gcloud cli, and sign in to your project. 
 # Make sure to use your PROJECT_ID value. 
 # The available models are gemini-2.5-flash-tts, gemini-2.5-flash-lite-preview-tts, and gemini-2.5-pro-tts. 
 # To parse the JSON output and use it directly see the last line of the command. 
 # Requires JQ and ffplay library to be installed. 
 PROJECT_ID 
 = 
 YOUR_PROJECT_ID 
curl  
-X  
POST  
 \ 
  
-H  
 "Authorization: Bearer 
 $( 
gcloud  
auth  
application-default  
print-access-token ) 
 " 
  
 \ 
  
-H  
 "x-goog-user-project: 
 $PROJECT_ID 
 " 
  
 \ 
  
-H  
 "Content-Type: application/json" 
  
 \ 
-d  
 '{ 
 "contents": { 
 "role": "user", 
 "parts": { "text": "Say the following in a curious way: OK, so... tell me about this [uhm] AI thing." } 
 }, 
 "generation_config": { 
 "speech_config": { 
 "language_code": "en-in", 
 "voice_config": { 
 "prebuilt_voice_config": { 
 "voice_name": "kore" 
 } 
 } 
 } 
 } 
 }' 
  
 \ 
  
https://aiplatform.googleapis.com/v1beta1/projects/ $PROJECT_ID 
/locations/us-central1/publishers/google/models/gemini-2.5-flash-tts:generateContent  
 \ 
  
 | 
  
jq  
-r  
 '.candidates[0].content.parts[0].inlineData.data' 
  
 \ 
  
 | 
  
base64  
-d  
 | 
  
ffmpeg  
-f  
s16le  
-ar  
24k  
-ac  
 1 
  
-i  
-  
output_speech.wav 

  from 
  
 google 
  
 import 
 genai 
 from 
  
 google.genai 
  
 import 
 types 
 import 
  
 wave 
 import 
  
 os 
 import 
  
 datetime 
 PROJECT_ID 
 = 
 os 
 . 
 getenv 
 ( 
 "GOOGLE_CLOUD_PROJECT" 
 ) 
 LOCATION 
 = 
 os 
 . 
 environ 
 . 
 get 
 ( 
 "GOOGLE_CLOUD_REGION" 
 , 
 "global" 
 ) 
 # Set up the wave file to save the output: 
 def 
  
 wave_file 
 ( 
 filename 
 , 
 pcm 
 , 
 channels 
 = 
 1 
 , 
 rate 
 = 
 24000 
 , 
 sample_width 
 = 
 2 
 ): 
 with 
 wave 
 . 
 open 
 ( 
 filename 
 , 
 "wb" 
 ) 
 as 
 wf 
 : 
 wf 
 . 
 setnchannels 
 ( 
 channels 
 ) 
 wf 
 . 
 setsampwidth 
 ( 
 sample_width 
 ) 
 wf 
 . 
 setframerate 
 ( 
 rate 
 ) 
 wf 
 . 
 writeframes 
 ( 
 pcm 
 ) 
 def 
  
 synthesize 
 ( 
 text 
 : 
 str 
 , 
 model 
 : 
 str 
 , 
 voice 
 : 
 str 
 , 
 locale 
 : 
 str 
 ): 
  
 """Synthesizes speech from the input text. 
 Args: 
 text: Text to synthesize. 
 model: gemini tts model name. gemini-2.5-flash-tts, gemini-2.5-flash-lite-preview-tts, and gemini-2.5-pro-tts 
 voice: voice name. Example: leda, kore. Refer to available voices 
 locale: locale name. Example: en-us. Refer to available locales. 
 """ 
 client 
 = 
 genai 
 . 
 Client 
 ( 
 vertexai 
 = 
 True 
 , 
 project 
 = 
 PROJECT_ID 
 , 
 location 
 = 
 LOCATION 
 ) 
 generate_content_config 
 = 
 types 
 . 
 GenerateContentConfig 
 ( 
 speech_config 
 = 
 types 
 . 
 SpeechConfig 
 ( 
 language_code 
 = 
 locale 
 , 
 voice_config 
 = 
 types 
 . 
 VoiceConfig 
 ( 
 prebuilt_voice_config 
 = 
 types 
 . 
 PrebuiltVoiceConfig 
 ( 
 voice_name 
 = 
 voice 
 , 
 ) 
 ) 
 ), 
 ) 
 request_start_time 
 = 
 datetime 
 . 
 datetime 
 . 
 now 
 () 
 is_first_chunk_received 
 = 
 False 
 final_audio_data 
 = 
 bytes 
 () 
 num_chunks_received 
 = 
 0 
 for 
 chunk 
 in 
 client 
 . 
 models 
 . 
 generate_content_stream 
 ( 
 model 
 = 
 model 
 , 
 contents 
 = 
 text 
 , 
 config 
 = 
 generate_content_config 
 , 
 ): 
 # just a simple progress indicator 
 num_chunks_received 
 += 
 1 
 print 
 ( 
 "." 
 , 
 end 
 = 
 "" 
 ) 
 if 
 num_chunks_received 
 % 
 40 
 == 
 0 
 : 
 print 
 ( 
 "" 
 ) 
 # measuring time to first audio 
 if 
 not 
 is_first_chunk_received 
 : 
 is_first_chunk_received 
 = 
 True 
 first_chunk_received_time 
 = 
 datetime 
 . 
 datetime 
 . 
 now 
 () 
 if 
 ( 
 chunk 
 . 
 candidates 
 is 
 None 
 or 
 not 
 chunk 
 . 
 candidates 
 or 
 chunk 
 . 
 candidates 
 [ 
 0 
 ] 
 . 
 content 
 is 
 None 
 or 
 not 
 chunk 
 . 
 candidates 
 [ 
 0 
 ] 
 . 
 content 
 . 
 parts 
 ): 
 continue 
 part 
 = 
 chunk 
 . 
 candidates 
 [ 
 0 
 ] 
 . 
 content 
 . 
 parts 
 [ 
 0 
 ] 
 if 
 part 
 . 
 inline_data 
 and 
 part 
 . 
 inline_data 
 . 
 data 
 : 
 # accumulating audio. In a web-server scenario, you would want to 
 # "emit" audio to the frontend as soon as it arrives. 
 # 
 # For example using flask socketio, you could do the following 
 # from flask_socketio import SocketIO, emit 
 # emit("audio", chunk.candidates[0].content.parts[0].inline_data.data) 
 # socketio.sleep(0) 
 final_audio_data 
 += 
 chunk 
 . 
 candidates 
 [ 
 0 
 ] 
 . 
 content 
 . 
 parts 
 [ 
 0 
 ] 
 . 
 inline_data 
 . 
 data 
 time_to_first_audio 
 = 
 first_chunk_received_time 
 - 
 request_start_time 
 time_to_completion 
 = 
 datetime 
 . 
 datetime 
 . 
 now 
 () 
 - 
 request_start_time 
 print 
 ( 
 " 
 \n 
 " 
 ) 
 print 
 ( 
 f 
 "Time to first audio: 
 { 
 time_to_first_audio 
 . 
 total_seconds 
 () 
 } 
 seconds" 
 ) 
 print 
 ( 
 f 
 "Time to completion: 
 { 
 time_to_completion 
 . 
 total_seconds 
 () 
 } 
 seconds" 
 ) 
 return 
 final_audio_data 
 audio_data 
 = 
 synthesize 
 ( 
 "Say the following in a curious way: Radio Bakery is a New York City gem, celebrated for its exceptional and creative baked goods. The pistachio croissant is often described as a delight with perfect sweetness. The rhubarb custard croissant is a lauded masterpiece of flaky pastry and tart filling. The brown butter corn cake stands out with its crisp edges and rich flavor. Despite the bustle, the staff consistently receives praise for being friendly and helpful." 
 , 
 "gemini-2.5-flash-tts" 
 , 
 "Kore" 
 , 
 "en-in" 
 ) 
 file_name 
 = 
 'output_speech.wav' 
 wave_file 
 ( 
 file_name 
 , 
 audio_data 
 ) 
 Audio 
 ( 
 "output_speech.wav" 
 ) 
 

  # Make sure to install gcloud cli, and sign in to your project. 
 # Make sure to use your PROJECT_ID value. 
 # The available models are gemini-2.5-flash-tts, gemini-2.5-flash-lite-preview-tts, and gemini-2.5-pro-tts. 
 # To parse the JSON output and use it directly see the last line of the command. 
 # Requires JQ and ffplay library to be installed. 
 PROJECT_ID 
 = 
 YOUR_PROJECT_ID 
curl  
-X  
POST  
 \ 
  
-H  
 "Authorization: Bearer 
 $( 
gcloud  
auth  
application-default  
print-access-token ) 
 " 
  
 \ 
  
-H  
 "x-goog-user-project: 
 $PROJECT_ID 
 " 
  
 \ 
  
-H  
 "Content-Type: application/json" 
  
 \ 
-d  
 '{ 
 "contents": { 
 "role": "user", 
 "parts": { "text": "Say the following as a conversation between friends: Sam: Hi Bob, how are you?\\nBob: I am doing well, and you?" } 
 }, 
 "generation_config": { 
 "speech_config": { 
 "language_code": "en-in", 
 "multi_speaker_voice_config": { 
 "speaker_voice_configs": [{ 
 "speaker": "Sam", 
 "voice_config": { 
 "prebuilt_voice_config": { 
 "voice_name": "Aoede" 
 } 
 } 
 },{ 
 "speaker": "Bob", 
 "voice_config": { 
 "prebuilt_voice_config": { 
 "voice_name": "Algieba" 
 } 
 } 
 }] 
 } 
 } 
 } 
 }' 
  
 \ 
  
https://aiplatform.googleapis.com/v1beta1/projects/ $PROJECT_ID 
/locations/us-central1/publishers/google/models/gemini-2.5-flash-tts:generateContent  
 \ 
  
 | 
  
jq  
-r  
 '.candidates[0].content.parts[0].inlineData.data' 
  
 \ 
  
 | 
  
base64  
-d  
 | 
  
ffmpeg  
-f  
s16le  
-ar  
24k  
-ac  
 1 
  
-i  
-  
output_speech.wav 

  # google-cloud-texttospeech minimum version 2.32.0 is required. 
 request 
 = 
 texttospeech 
 . 
 SynthesizeSpeechRequest 
 ( 
 input 
 = 
 synthesis_input 
 , 
 voice 
 = 
 voice 
 , 
 audio_config 
 = 
 audio_config 
 , 
 advanced_voice_options 
 = 
 texttospeech 
 . 
 AdvancedVoiceOptions 
 ( 
 relax_safety_filters 
 = 
 True 
 , 
 ), 
 ) 
 # Perform the text-to-speech request on the text input with the selected 
 # voice parameters and audio file type. 
 response 
 = 
 client 
 . 
 synthesize_speech 
 ( 
 request 
 = 
 request 
 ) 
 

  PROJECT_ID 
 = 
 YOUR_PROJECT_ID 
curl  
-X  
POST  
 \ 
  
-H  
 "Authorization: Bearer 
 $( 
gcloud  
auth  
application-default  
print-access-token ) 
 " 
  
 \ 
  
-H  
 "x-goog-user-project: 
 $PROJECT_ID 
 " 
  
 \ 
  
-H  
 "Content-Type: application/json" 
  
 \ 
-d  
 '{ 
 "input": { 
 "prompt": "Say the following in a curious way", 
 "text": "I am saying something that would otherwise be blocked by Gemini TTS." 
 }, 
 "voice": { 
 "languageCode": "en-us", 
 "name": "Kore", 
 "modelName": "gemini-2.5-flash-tts" 
 }, 
 "audioConfig": { 
 "audioEncoding": "LINEAR16" 
 }, 
 "advancedVoiceOptions": { 
 "relaxSafetyFilters": true 
 } 
 }' 
  
 \ 
  
 "https://texttospeech.googleapis.com/v1/text:synthesize" 
  
 \ 
  
 | 
  
jq  
-r  
 '.audioContent' 
  
 | 
  
base64  
-d  
 | 
  
ffplay  
-  
-autoexit