Speech-to-Text / Text-to-Speech / Translation API Complete Guide

GCP's voice and language AI is built around three pillars: Speech-to-Text, Text-to-Speech, and Translation API. Chirp 2 (USM), WaveNet / Journey TTS, and Translation V3 deliver industry-leading accuracy and multilingual coverage, making them the standard toolkit for voice and translation features in global apps.

Speech-to-Text (STT) Models

STT Sample (Real-Time)

from google.cloud import speech_v2 as speech

client = speech.SpeechClient()

config = speech.RecognitionConfig(
    auto_decoding_config=speech.AutoDetectDecodingConfig(),
    language_codes=["ja-JP"],
    model="chirp_2",
)
streaming_config = speech.StreamingRecognitionConfig(config=config)
config_request = speech.StreamingRecognizeRequest(
    recognizer=f"projects/PROJECT/locations/asia-northeast1/recognizers/_",
    streaming_config=streaming_config,
)

def audio_generator():
    yield config_request
    for chunk in audio_stream:  # マイク等から
        yield speech.StreamingRecognizeRequest(audio=chunk)

for response in client.streaming_recognize(requests=audio_generator()):
    for result in response.results:
        print(result.alternatives[0].transcript)

Text-to-Speech (TTS) Voice Types

TTS Sample

from google.cloud import texttospeech

client = texttospeech.TextToSpeechClient()
text = "こんにちは、世界"

synthesis_input = texttospeech.SynthesisInput(text=text)
voice = texttospeech.VoiceSelectionParams(
    language_code="ja-JP",
    name="ja-JP-Neural2-B",
)
audio_config = texttospeech.AudioConfig(
    audio_encoding=texttospeech.AudioEncoding.MP3,
)

response = client.synthesize_speech(
    input=synthesis_input, voice=voice, audio_config=audio_config
)
with open("output.mp3", "wb") as out:
    out.write(response.audio_content)

Translation API V3 (Advanced)

• Translation: 100+ language pairs, neural machine translation
• Batch Translation: Bulk file processing (via GCS)
• Glossary: Controlled translation of specialized terms and proper nouns
• AutoML Translation: Industry-specific custom models
• Adaptive Translation: Domain-adaptive translation (Adaptive MT)
• Document Translation: Format-preserving translation for PDF / Word / PPT

Pricing (Translation, 2026)

Comparison with Other Clouds

Typical Use Cases

• Automated meeting minutes (STT + Gemini for summarization)
• Call center transcription + sentiment analysis
• Video subtitle generation (STT + Translation for multilingual subtitles)
• Multilingual IVR (STT + Translation + TTS)
• e-Learning audio materials (TTS Studio)
• Multilingual e-commerce product descriptions (Translation + Glossary)
• Voice navigation (TTS Journey)

※ Google Cloud is a trademark of Google LLC. For the latest information, see the official Speech-to-Text docs.