AI SPEAKER SEPARATION
Turn one mixed podcast, interview, or recording into individual tracks, with speakers split automatically and kept time-aligned.
Separation workspace
Upload an audio or video recording for AI speaker separation, choose automatic or manual detection, and get speakers split into individual, time-aligned tracks ready to preview and download.
SEPARATION TOOL
Sign in to upload a recording
Audio: WAV, MP3, M4A, AAC, FLAC, OGG, OPUS, WMA, Speex. Video: MP4, AVI, MOV, MKV, WebM. Uploads up to 1 GB. Compatible videos upload only their original audio track, without re-encoding.
录音会保持私密，并与完成的音轨一起保存 30 天。
Listen closely
Switch between the original mix and isolated speaker tracks without losing your place.
How it works
One continuous signal path turns a mixed recording into organized, editor-ready tracks.
STEP 01
Choose a mixed podcast, interview, meeting, or conversation.
STEP 02
AI identifies the voices and creates a time-aligned track for each detected speaker.
STEP 03
Listen to every speaker and download each separated track as an individual WAV file.
Features
From speaker detection to secure delivery, every part of the workflow stays focused on usable audio.
CORE OUTPUT
AI speaker separation gives each detected speaker an independent audio track.
Every track starts at the same point and stays aligned with the original recording.
Let Seply detect speakers automatically, or enter the expected number yourself.
Use Seply with podcasts, interviews, meetings, panels, and other multi-speaker recordings.
Download individual, time-aligned WAV files for your editing workflow.
Each Seply feature has its own credit rate. See the estimated cost before processing begins.
Separation vs diarization
Speaker diarization
00:00-00:12 Speaker 1
00:12-00:19 Speaker 2
00:18-00:24 Speaker 1 + Speaker 2
Speaker diarization tells you who spoke when.
Speaker separation
AI speaker separation creates an actual track for each speaker, so you can edit, mute, enhance, or reuse every voice independently.
Use cases
Keep every voice editable, whether the final destination is an episode, research archive, meeting record, or video timeline.
Separate hosts and guests so you can adjust levels, remove mistakes, and edit each voice independently.
Isolate interviewers and participants for easier review, analysis, and speaker-specific workflows.
Turn one mixed discussion into separate tracks for review, archiving, or post-production.
Prepare independent dialogue tracks before placing them back into your video editing timeline.
价格
选择月度套餐，或需要时单独购买积分。
一次性 30 积分
一次性试用，30 天内有效，无需信用卡。
每月 200 积分
适合偶尔处理采访和短录音。
最受欢迎
每月 600 积分
适合持续制作播客和采访。
每月 1500 积分
适合每周处理音频的团队。
无需订阅。积分自购买之日起 12 个月内有效。
每积分 $0.080
200 积分
$16
每积分 $0.065
600 积分
$39
每积分 $0.059
1500 积分
$89
所有套餐均包含
Speaker Separation 的费率为每分钟 10 积分，按 6 秒递增计费，最低扣除 1 积分。1 分 01 秒的录音扣除 11 积分。
Prices are in USD. Monthly plans renew automatically until canceled from Billing. See the Refund Policy.
常见问题
了解文件、输出、积分和说话人分离的相关规则。
Seply 可将一段混合录音分离成每位说话人各自独立且时间对齐的音轨。
不是。说话人日志标注谁在何时说话；说话人分离会为每位说话人生成实际的独立音轨。
Seply 面向多人录音，也支持包含重叠对话的场景。实际效果会受到噪声、回声、录音质量、声音相似度和重叠程度影响。
Seply 支持上传区列出的音频和视频格式，包括 WAV、MP3、M4A、AAC、FLAC、OGG、OPUS、WMA、Speex、MP4、AVI、MOV、MKV 和 WebM。
每位检测到的说话人都会获得一条独立且时间对齐的音频文件。
积分代表 Seply 平台内可消费的处理能力，每项功能都有独立费率。Speaker Separation 的费率为每分钟 10 积分，按 6 秒递增计费且最低扣除 1 积分，因此 1 分 01 秒扣除 11 积分。最终以服务端解析的媒体时长为准。
月度套餐积分自发放起 30 天后过期；一次性积分包自购买起 12 个自然月后过期。取消订阅不会提前清空已发放积分。
任务明确失败后，预扣积分会自动退回原积分批次，并保留原到期时间。
对于新任务，原始录音和完成的说话人音轨会从任务完成起私密保存 30 天。在此期间可在历史记录中试听并下载分离音轨；媒体过期后仍会保留有限的任务元数据。
购买后 7 个自然日内，未使用的已购积分可以申请退款。已使用积分和已完成任务原则上不退款；重复扣款、无法交付及法律要求的权利除外。