Oracle Cloud Infrastructure Speech
oracle.com
What it is
Oracle Cloud Infrastructure Speech transcribes submitted audio to text and synthesizes speech from text. Access is available through the OCI Console, REST APIs, CLIs, SDKs, and OCI Data Science notebook sessions. File-based asynchronous transcription is supported; real-time transcription is in limited availability. Transcription output defaults to JSON, with SRT also available at no additional cost. Tools include profanity filtering, punctuation, diarization, and word-level and full-transcription confidence scores. Supported audio includes single-channel 16-bit PCM WAV at 16 kHz, along with AAC, AC3, AMR, AU, FLAC, M4A, MKV, MP3, MP4, OGA, OGG, and WEBM. Oracle lists nine languages and says automatic language detection is not currently available. The first five transcription hours per month per tenancy are free; additional transcription or voice-synthesis hours cost $0.50 per hour, with usage aggregated in seconds. There are no setup charges, minimum commitments, or hardware requirements. Oracle says it retains no information from submitted files. Audio quality, accents, background noise, language switching, and simultaneous speakers can affect transcription quality.
Spec sheet
- What it does
- OCI Speech transcribes audio to text asynchronously or in real time and synthesizes speech from text.Read 12 Oct 2026 · oracle.com
- Platforms
- web,apiRead 20 Sep 2026 · oracle.com
- Free plan
- YesRead 20 Sep 2026 · oracle.com
- Privacy
- Oracle says it only transcribes submitted content and keeps no information from the file.Read 12 Oct 2026 · oracle.com
- Support
- Oracle’s Speech documentation links to support requests and service-request creation.Read 12 Oct 2026 · docs.oracle.com
- Access
- The service is available through the OCI Console, REST APIs, CLIs, SDKs, and OCI Data Science notebook sessions.Read 12 Oct 2026 · oracle.com
- API access
- YesRead 20 Sep 2026 · oracle.com
- Audio formats
- Supported formats include single-channel 16-bit PCM WAV at 16 kHz and AAC, AC3, AMR, AU, FLAC, M4A, MKV, MP3, MP4, OGA, OGG, and WEBM.Read 12 Oct 2026 · oracle.com
- Audio quality
- Oracle states transcription quality depends on input audio quality and may be affected by accents, background noise, language switching, and simultaneous speakers.Read 12 Oct 2026 · oracle.com
- Commitment
- OCI Speech has no setup charges, minimum service commitments, or hardware requirement.Read 12 Oct 2026 · oracle.com
- Export formats
- MP3,PCM,OGG,JSONRead 20 Sep 2026 · oracle.com
- Language detection
- Automatic language detection is not currently available, according to Oracle’s FAQ.Read 12 Oct 2026 · oracle.com
- Languages
- 9Read 20 Sep 2026 · oracle.com
- Languages
- Pretrained models support English, Spanish, Portuguese, German, French, Italian, and Hindi; asynchronous file transcription with the OpenAI Whisper model supports 57+ languages.Read 12 Oct 2026 · oracle.com
- Maximum input
- 10000Read 20 Sep 2026 · oracle.com
- Output formats
- Transcription output is JSON by default, with SRT available at no additional cost.Read 12 Oct 2026 · oracle.com
- Pricing
- Oracle charges $0.50 per transcription or voice-synthesis hour, measures aggregated usage in seconds, and includes five free transcription hours each month per tenancy.Read 12 Oct 2026 · oracle.com
- Transcription modes
- File-based asynchronous transcription is supported, while real-time transcription is in limited availability.Read 12 Oct 2026 · oracle.com
- Transcription tools
- The service offers profanity filtering, word-level and full-transcription confidence scores, punctuation, and diarization.Read 12 Oct 2026 · oracle.com
- Use cases
- Oracle describes use for customer-call transcription, subtitling, and generating searchable metadata for media archives.Read 12 Oct 2026 · docs.oracle.com
Plans and prices
Strong points and limits
Strong points
- Five transcription hours per month per tenancy are free
- Supports asynchronous file transcription
- SRT is available at no additional cost
- No setup charge or minimum commitment
- Offers confidence scores and diarization
Limits
- Real-time transcription is in limited availability
- Automatic language detection is unavailable
- Additional use costs $0.50 per hour
- Audio conditions can affect transcription quality
Who it suits
It suits users building transcription or speech-synthesis workflows through OCI interfaces or APIs. Call transcription, subtitling, and searchable media metadata are among the use cases Oracle describes.
Verdict
OCI Speech offers transcription and speech synthesis with API and console access, plus several transcription controls. Account for the five-hour monthly allowance, usage charge, and audio-quality dependencies when assessing fit.
Where Specifiction places it
- Text-to-speech tools#17 of 2339.0
- Text-to-Speech Software#55 of 2238.5