AI text-to-speech voices, side by side
The same scripts, read by each service. Names are hidden at first so you can judge by ear; the voice, model and settings are listed under Test conditions.
AI-generated audioEvery sample on this page is synthetic speech made with each service's text-to-speech.
Audio: Amazon Polly , Azure AI Speech , Cartesia , CoeFont , ElevenLabs(affiliate) , Google Cloud Text-to-Speech , VOICEVOX (in no particular order)
Japanese, news
今日は10月16日です。ひがし町図書館は、11月1日から開館時間を延ばします。新しい時間は午前8時30分から午後7時30分までで、一日あたり2時間長くなります。貸出カードは年間450円、1人5冊まで、期間は2週間です。ご質問は窓口かFAQページへどうぞ。夜の静かな学習室も使えます。お近くの方は、ぜひ寄ってみませんか。
- Amazon Polly AI-generated audio
- Azure AI Speech AI-generated audio
- Cartesia AI-generated audio
-
-
- Google Cloud Text-to-Speech AI-generated audio
-
English, news
The public library in Marlowe Green will extend its opening hours starting March 3. Doors will open at 8:30 a.m. and close at 7 p.m., which adds about 2.5 hours each week. A library card for visitors from outside the district costs $4.50 per year. The extra time will be used for quiet study, a children's story hour, and free Wi-Fi. Questions can be sent by email, or asked in person at the main desk.
- Amazon Polly AI-generated audio
- Azure AI Speech AI-generated audio
- Cartesia AI-generated audio
-
- Google Cloud Text-to-Speech AI-generated audio
Liked a voice?
Each service lets you try it on its free plan. Go straight from here.
Test conditions
Every service read the same scripts under the same conditions. Each service name links to that service's own site.
| Method | each service's API, or its web app where noted below |
|---|---|
| Settings | each service's default settings, except where noted per service in notes |
| Format | mp3 64 kbps mono 44.1 kHz; loudness normalized to -16 LUFS integrated (ffmpeg loudnorm, two-pass, true peak -1.5 dBTP); metadata stripped |
| Notes | Each sample is the result of a single generation, used as is (converted to the common format). - amazon-polly (en): voice Joanna; model neural; settings defaults; generated via API - amazon-polly (ja): voice Kazuha; model neural; settings defaults; generated via API - azure-speech (en): voice en-US-AvaNeural; model Neural (Standard); settings defaults; generated via API - azure-speech (ja): voice ja-JP-NanamiNeural; model Neural (Standard); settings defaults; generated via API - cartesia (en): voice Skylar [id db6b0ed5-d5d3-463d-ae85-518a07d3c2b4]; model sonic-3.6; settings defaults; generated via API - cartesia (ja): voice Yuzuki [id cb21b4af-0e46-44f7-9128-51be7f898311]; model sonic-3.6; settings defaults; generated via API - coefont (ja): voice 林 梨沙; model CoeFont; settings defaults; generated by hand in the service's own app and imported; credit Voiced by https://CoeFont.cloud - elevenlabs (en): voice Sarah [id EXAVITQu4vr4xnSDxMaL]; model eleven_v4; settings defaults; generated via API; credit elevenlabs.io - elevenlabs (ja): voice Yui; model eleven_v4; settings defaults; generated by hand in the service's own app and imported; credit elevenlabs.io - google-cloud-tts (en): voice en-US-Chirp3-HD-Achernar; model Chirp 3: HD; settings defaults; generated via API - google-cloud-tts (ja): voice ja-JP-Chirp3-HD-Achernar; model Chirp 3: HD; settings defaults; generated via API - voicevox (ja): voice 四国めたん（ノーマル） [id 2]; model VOICEVOX ENGINE 0.25.2; settings defaults; generated via API; credit VOICEVOX:四国めたん |
| Credits |
|
| Generated on | |
| Scripts |
|
Voices and models
| Script | Service | Voice | Model |
|---|---|---|---|
| Japanese, news | Amazon Polly | Kazuha | neural |
| Japanese, news | Azure AI Speech | ja-JP-NanamiNeural | Neural (Standard) |
| Japanese, news | Cartesia | Yuzuki | sonic-3.6 |
| Japanese, news | CoeFont | 林 梨沙 | CoeFont |
| Japanese, news | ElevenLabs | Yui | eleven_v4 |
| Japanese, news | Google Cloud Text-to-Speech | ja-JP-Chirp3-HD-Achernar | Chirp 3: HD |
| Japanese, news | VOICEVOX | 四国めたん（ノーマル） | VOICEVOX ENGINE 0.25.2 |
| English, news | Amazon Polly | Joanna | neural |
| English, news | Azure AI Speech | en-US-AvaNeural | Neural (Standard) |
| English, news | Cartesia | Skylar | sonic-3.6 |
| English, news | ElevenLabs | Sarah | eleven_v4 |
| English, news | Google Cloud Text-to-Speech | en-US-Chirp3-HD-Achernar | Chirp 3: HD |
This page has affiliate links, marked “(affiliate)”. If you sign up or buy through them, TierDiff may earn a commission. It never affects the numbers or the order. See the disclosure.