Speech 2.8 Turbo
Speech 2.8 TurboMiniMax Speech 2.8 Turbo delivers lifelike, studio-clear AI voices with native sound tags, 10-second voice cloning, and stronger cross-lingual speech worldwide.
Model details
Example usage
MiniMax Speech 2.8 Turbo is an advanced text-to-speech model designed to produce lifelike, expressive voices. Native sound tags add natural breaths, pauses, hesitations, laughter, and other vocal reactions, creating speech with more authentic rhythm and emotion.
The model also offers high-fidelity voice cloning from a 10-second sample, capturing details such as vocal texture, breathiness, and speaking pace. Studio-grade audio clarity reduces background noise and digital artifacts, while improved cross-lingual performance delivers more natural pronunciation across languages.
See developer guide.
1import requests
2
3url = os.getenv("MODEL_URL")
4
5payload = {
6 "model": "speech-2.8-hd",
7 "text": "Omg(sighs), the real danger is not that computers start thinking like people, but that people start thinking like computers. Computers can only help us with simple tasks.",
8 "stream": False,
9 "voice_setting": {
10 "voice_id": "English_expressive_narrator",
11 "speed": 1,
12 "vol": 1,
13 "pitch": 0
14 },
15 "audio_setting": {
16 "sample_rate": 32000,
17 "bitrate": 128000,
18 "format": "mp3",
19 "channel": 1
20 },
21 "pronunciation_dict": { "tone": ["Omg/Oh my god"] },
22 "language_boost": "auto",
23 "voice_modify": {
24 "pitch": 0,
25 "intensity": 0,
26 "timbre": 0,
27 "sound_effects": "spacious_echo"
28 },
29 "output_format": "hex"
30}
31headers = {
32 "Content-Type": "application/json",
33 "Authorization": f"Bearer {os.getenv('API_KEY')}"
34}
35
36response = requests.post(url, json=payload, headers=headers)
37
38print(response.text)1{
2 "data": {
3 "audio": "<hex encoded audio>",
4 "status": 2
5 },
6 "extra_info": {
7 "audio_length": 11124,
8 "audio_sample_rate": 32000,
9 "audio_size": 179926,
10 "bitrate": 128000,
11 "word_count": 163,
12 "invisible_character_ratio": 0,
13 "usage_characters": 163,
14 "audio_format": "mp3",
15 "audio_channel": 1
16 },
17 "trace_id": "01b8bf9bb7433cc75c18eee6cfa8fe21",
18 "base_resp": {
19 "status_code": 0,
20 "status_msg": "success"
21 }
22}
