Testing Meta’s SAM Audio: Isolating Voices and Instruments

Matthew Berman demonstrates prompt-based sound separation and effects in a Meta-partnered video.

MBMatthew Berman登壇者 · Forward Future
チャンネル
Matthew Berman
公開日
2026年1月6日
動画
5:13
読了時間
1 分
スライド
2
原文
英語
YouTube で開く
スクロールすると動画は右下に小さくなって再生を続けます。字幕は閲覧言語に合わせて切り替わります。
閲覧言語2 言語で読めます
この記事はまだ日本語に翻訳されていないため、英語の原文を表示しています。
要約

30 秒でわかる

Matthew Berman tests Meta’s SAM Audio playground with game footage, a café recording and music. He separates a target sound from the remaining audio, applies effects and describes possible uses. The video was made in partnership with Meta; hearing-aid applications are a future possibility rather than a demonstrated product.

  1. 1Meta partnershipBerman discloses that he partnered with Meta for the demonstration.
  2. 2Separate sound tracksA prompt selects a woman’s sound in a game clip, and the interface offers isolated and remaining tracks.
  3. 3Café soundsHe tries voice, footsteps and utensils in a noisy restaurant video.
  4. 4Music and effectsHe isolates guitar, applies effects and describes downloading tracks.

ポイント

  1. A text prompt can identify a sound to isolate from a video or audio clip.
  2. The playground presents the original, isolated sound and remaining audio as separate tracks.
  3. Berman tests speech, footsteps, utensils and guitar, then applies effects.
  4. The video is a Meta partnership; the examples are demonstrations rather than a comprehensive evaluation.
全文

チャプターごとに整えた全文を読む

文字起こしのみ
01

Sound separation in video

Berman introduces the Meta partnership and isolates a sound in game footage.

Meta’s SAM Audio can isolate sounds from audio or video using a prompt. In this video, made in partnership with Meta, I demonstrate its playground and show how the separated tracks can be used.

I upload a Tomb Raider clip and enter “woman.” The interface produces the original audio, the isolated sound and the remaining sound. I play the isolated track and then the remainder; the result sounds unusual to me, but the selected voice is separated in this example.

02

Voices and background sounds

He separates speech and other sounds from a café clip, then applies a voice effect.

Next I use a café video where a woman speaks on the phone amid background noise. I enter “voice” and listen to the isolated speech and the remainder. I also try “footsteps” and “utensils” to select other sounds from the same scene.

Isolated and remaining audio
スライド 1 / 2Isolated and remaining audioThe café example shows separate tracks and controls for applying effects.

The playground offers effects on the isolated audio or the remainder. I apply a studio-sound effect to the voice and adjust its level; other options include reverb and more stylized effects.

03

Music, effects and export

He tests instrument isolation and effects, then discusses possible audio-production uses.

For a music clip I isolate guitar, listen to the remaining track and try concert-hall and underwater effects.

Effects on a music clip
スライド 2 / 2Effects on a music clipThe interface displays the original, isolated and remaining tracks with an underwater effect selected.

The interface lets me download tracks individually. I see potential uses in cleanup and mixing; a hearing aid using similar technology is an idea for the future, not something demonstrated here.

同じチャンネルの記事

スライド 6 枚25:50EN
MBMatthew Berman· 2026年10月24日Kimi K3.1 Pro Tested: Strong 3D Demos, Uneven Results and Competitive PricingMatthew Berman · Forward FutureMatthew Berman tests the open-weight model on coding, 3D rendering and visual explanations, then compares benchmark results and prices.Open-weight modelsBenchmarksAI cost & performance+1
スライド 2 枚32:20EN
MBMatthew Berman· 2026年10月23日Who Is Local AI For? The Costs, Trade-offs and a Hybrid FutureMatthew Berman · Forward FutureMatthew Berman weighs data control and independence against hardware cost, setup complexity and inference speed.Local inferenceOpen-weight modelsAI cost & performance+1
45:08EN
MBMatthew Berman· 2026年10月21日Sam Altman on AI Discovery, Consciousness, Compute and a Human-Centered FutureMatthew Berman, Sam Altman · Forward FutureIn a conversation with Matthew Berman, Altman discusses scientific progress, personal assistants, recursive self-improvement and the importance of human agency.AI for sciencePersonal agentsConsciousness+2
0:32EN
MBMatthew Berman· 2026年10月21日Sam Altman on the Continuing Demand for Software WorkSam Altman, Matthew Berman · OpenAIIn a short interview excerpt, Altman says demand for software may be larger than expected even as AI changes how it is made.Future of WorkSoftware engineeringコーディングエージェント
スライド 5 枚12:14EN
MBMatthew Berman· 2026年10月14日Gemini 4 Argon Tested: 3D Demos, a Spreadsheet Prototype and Benchmark ResultsMatthew Berman · Forward FutureMatthew Berman tests Google’s model on games and visual projects, then compares reported coding and agent benchmarks.GeminiBenchmarksコーディングエージェント+1
スライド 5 枚10:36EN
MBMatthew Berman· 2026年10月13日Meta Muse Wonder Tested: Strong Benchmarks, Mixed Demos and Quota FrictionMatthew Berman · Forward FutureMatthew Berman tests Meta’s models on browser demos and reviews their published benchmark scores, availability and cost.モデルと学習BenchmarksAI cost & performance+1