James had been producing a weekly interview podcast for three years before he tried Descript. The promise was compelling: edit your audio by editing the transcript like a document, remove filler words in bulk, clean up silences automatically, and generate show notes from the same interface. He paid for the Pro plan at $24/month and spent a weekend learning the tool. The transcript-based editing was real -- it worked as described, and cutting sections of an interview by selecting and deleting text in the transcript was faster than timeline editing in Audacity for interview content where the cuts were driven by what was said rather than where peaks appeared in the waveform.
Three months in, he was using Descript for about twenty percent of its capability. The transcript-based editing for interview shows. The overdub voice cloning, which he tried once and found ethically uncomfortable to use for anything beyond fixing a mispronounced word. The studio sound processing for background noise. Most of the other features -- screen recording, multi-track mixing, video editing -- he never touched. The monthly payment continued. Then his interview guest for a particularly important episode was a medical researcher with a thick Scottish accent and a tendency to use acronyms from her specific research field. The AI transcription was sixty-three percent accurate. He spent three hours manually correcting the transcript before editing could begin.
That experience prompted a more systematic look at what he actually needed: accurate transcription, filler word removal, and show notes generation. Descript covered two of those reliably. He added Castmagic for show notes generation, where the AI output was dramatically more useful than doing it manually, and kept Descript for editing. But the experience of finding better tools for specific parts of the workflow made him look at the category more carefully.
"Descript's transcript-editing model is a genuine innovation. The challenge is that innovation in transcription accuracy, content repurposing, and meeting notes has happened in tools that do not try to replace your entire production workflow."
Why People Look for Descript Alternatives
Descript's value proposition is coherent: it wants to be the single tool for recording, transcribing, editing, and publishing audio and video content. For creators who buy into that vision fully, Descript is a capable all-in-one. The reasons people look elsewhere cluster around a few consistent themes.
AI transcription accuracy is not universal. Descript's AI transcription performs well on clear audio with standard American or British accents and general vocabulary. It degrades on accents, technical terminology, multiple overlapping speakers, and poor audio quality. For creators working in these conditions -- international guests, academic or medical subject matter, noisy recording environments -- the accuracy is often insufficient, requiring significant manual correction that eliminates the time savings the transcript-based editing is supposed to provide.
The learning curve is real. Descript inverts the traditional audio/video editing paradigm. Editors trained in timeline-based tools like Adobe Premiere, Final Cut, or Audacity find that Descript requires relearning fundamental editing concepts. The transcript-as-interface model is powerful once internalized but takes genuine investment to reach fluency. Some users never feel fully comfortable with it.
The cost has risen relative to alternatives. At $24/month for the Pro plan, Descript is not expensive in absolute terms. But the transcription-specific market now includes Otter.ai at $17/month, Sonix at $22/month, and Fireflies.ai at $10/month, all of which handle the transcription use case without requiring adoption of an unfamiliar editing paradigm. For creators who only want transcription and not Descript's editing workflow, the alternatives are cheaper.
Overdub voice cloning raises ethical questions. Descript's overdub feature allows you to record a voice model and then generate new spoken audio from text using that voice -- useful for correcting mistakes without re-recording. The feature requires consent verification, but the technology's existence raises questions about consent for interview subjects, about editorial integrity, and about the appropriate use of AI-generated voice in journalistic or documentary contexts. Some creators avoid the tool partly to avoid the ethical territory the feature occupies.
Otter.ai
Otter.ai is a real-time meeting transcription tool that integrates with calendar systems and video conferencing platforms to automatically join and transcribe meetings.
Features: Real-time transcription with speaker identification, automatic meeting joining from Google Calendar and Outlook calendar integration, Zoom, Google Meet, and Microsoft Teams integration via OtterPilot (bot participant), AI-generated meeting summaries with action items and key points, shared meeting notebooks where team members can add comments and highlights, search across all transcripts, export to PDF, DOCX, and SRT, integration with Slack and Salesforce for pushing meeting summaries.
Pricing: Free: 300 transcription minutes/month, 30-minute limit per session. Pro $16.99/month: 1,200 minutes/month, AI chat with transcripts, export. Business $30/user/month: 6,000 minutes, admin controls, advanced features. Enterprise: custom.
Pros vs Descript: Real-time transcription during meetings is a capability Descript does not have. The meeting auto-join feature means transcription happens without any manual workflow. Simpler interface for the core transcription use case. Lower cost for transcription-only needs.
Cons vs Descript: No video editing. No filler word removal or audio processing. Accuracy is competitive but not best-in-class for difficult audio. The transcript is not connected to editable audio.
Best for: Teams and professionals who need automatic meeting transcription and notes with minimal setup, particularly those on Zoom, Google Meet, and Microsoft Teams.
Whisper / OpenAI
Whisper is OpenAI's open-source speech recognition model, released publicly in 2022 and now considered one of the most accurate AI transcription models available -- at zero cost for local use.
Features: Multi-language transcription and translation across 97 languages, multiple model sizes (tiny, base, small, medium, large) trading accuracy for processing speed and hardware requirements, speaker diarization available with additional open-source tools, SRT and VTT subtitle export, timestamps at word or sentence level, runs completely locally with no data sent to any server. The Large model achieves accuracy competitive with or exceeding paid services on standard audio.
Pricing: Free (open-source). OpenAI's Whisper API for cloud-based use is $0.006/minute. Third-party apps using Whisper charge their own prices.
Pros vs Descript: Zero cost for local use. No usage limits, no subscription, no data sent externally. Accuracy is among the best available for AI transcription. The open-source nature enables custom integrations and self-hosted deployment for privacy-sensitive content.
Cons vs Descript: No GUI -- requires command-line knowledge or a third-party app. No editing tools. No filler word removal. Setup complexity is a barrier for non-technical users. Processing speed on local hardware can be slow for long files without a capable GPU.
Best for: Technical users, developers, and privacy-conscious organizations that want high-accuracy AI transcription without cost or data privacy concerns. Access via user-friendly apps like MacWhisper (macOS) for non-technical users.
Rev
Rev provides both human transcription by professional transcriptionists and AI transcription, covering the accuracy spectrum from "good enough" to "broadcast ready."
Features: Human transcription at $1.50/minute with 99%+ accuracy, 24-hour turnaround for standard orders. AI transcription at $0.25/minute with faster turnaround. Captions and subtitles for video in SRT, VTT, and other formats. Translation into 45+ languages. Global English transcription handling non-native English speakers. Timestamped transcripts for video alignment. HIPAA compliance for healthcare content. REV API for integrating transcription into production workflows.
Pricing: Human transcription $1.50/minute. AI transcription $0.25/minute. Rush delivery available at premium. Subscription plans for high-volume users.
Pros vs Descript: Human transcription accuracy is unmatched for difficult audio -- accents, technical terminology, multiple speakers, poor recording quality. The hybrid model (AI transcription with optional human review) is practical for balancing cost and accuracy. HIPAA compliance for healthcare use.
Cons vs Descript: Per-minute pricing adds up for high-volume use. No editing tools -- Rev delivers a transcript file, not an editing environment. No meeting integration.
Best for: Journalists, lawyers, academics, and media producers who need accurate transcripts for archival, legal, or broadcast use, particularly content with difficult audio conditions where AI accuracy is insufficient.
Sonix
Sonix is a subscription-based AI transcription platform targeting media workflows with a built-in transcript editor, 35+ language support, and translation capabilities.
Features: Automated transcription with speaker identification, in-browser transcript editor for correcting AI errors, 35+ languages for transcription and translation within the same editor, team collaboration with shared transcript libraries, SRT/VTT/SCC caption export for video platforms, transcript search across all uploaded files, automated summaries and topic detection, custom vocabulary to improve accuracy on domain-specific terms, Zoom integration, integration with Adobe Premiere, Final Cut, and Avid via subtitle export.
Pricing: Pay-as-you-go $10/hour of audio. Premium subscription $22/month: unlimited transcription at discounted rates. Enterprise pricing for large organizations.
Pros vs Descript: Multi-language transcription and translation is a significant advantage for international content. The transcript editor is well-designed for correcting AI output. Per-hour pay-as-you-go pricing is cost-effective for irregular use. Good integration with professional video editing software via export formats.
Cons vs Descript: No video editing. No filler word removal or audio processing tools. No meeting transcription. The transcript is not connected to editable audio or video tracks.
Best for: Media producers, documentary filmmakers, journalists, and academic researchers who need high-quality transcription in multiple languages with professional export formats for subtitle and closed caption workflows.
Trint
Trint is a journalist-focused transcription platform with collaboration tools, in-article text search, and a workflow designed for broadcast and newsroom environments.
Features: AI transcription with in-browser editor, story builder that combines transcript segments into an article or script, search within all transcripts for research across a library of recordings, team collaboration with shared workspaces and commenting, highlight and tag system for organizing interview material, integration with Adobe Premiere, Final Cut, and Avid for subtitle export, 40+ language support, real-time collaboration on transcripts, compliance with GDPR and enterprise security standards.
Pricing: Individual $52/month: unlimited transcription, 3 concurrent files. Advanced $60/month: 10 concurrent files, team features. Enterprise $80+/month: custom seats, SSO, compliance.
Pros vs Descript: The story builder and research tools designed for journalists make it the strongest option for newsroom and documentary workflows where transcripts serve as raw material for reporting rather than just editing guides. Search across a library of recordings is genuinely valuable for long-form research projects. The brand has established trust in broadcast and news organizations.
Cons vs Descript: More expensive than most alternatives for individual users. No filler word removal or audio processing. No meeting transcription. The workflow is optimized for newsroom use cases rather than podcast or YouTube production.
Best for: Journalists, documentary producers, broadcast media teams, and research organizations that treat interview transcripts as source material for reporting and need to search, cross-reference, and organize across large libraries of recordings.
Castmagic
Castmagic is an AI content repurposing tool that takes audio or video recordings and generates multiple content formats automatically -- show notes, social posts, email newsletters, blog drafts, and more.
Features: Transcript generation from audio and video uploads, AI-generated show notes with chapter markers, social media posts (LinkedIn, Twitter/X, Instagram) with appropriate formatting and length, email newsletter drafts, blog article outlines and drafts, guest quotes and pull quotes, podcast questions for follow-up episodes, custom prompt templates for generating any text format from the transcript, support for video content in addition to audio. The core differentiator is that Castmagic treats the transcript as input for downstream content generation rather than the end product.
Pricing: Starter $39/month: 160 upload minutes. Pro $59/month: 600 minutes. Agency $99/month: 1,200 minutes, team features.
Pros vs Descript: Content repurposing output is more useful and more automated than anything Descript generates. For podcasters and video creators who spend significant time creating show notes, social posts, and newsletters from their recordings, Castmagic eliminates most of that manual work. The AI-generated output quality is practical -- it requires editing but is substantially better than starting from scratch.
Cons vs Descript: No editing tools -- Castmagic delivers text, not an editing environment. The output is a starting point that requires human editing and judgment. Per-upload pricing limits are a constraint for high-volume creators.
Best for: Podcasters, YouTubers, and content creators who want to maximize the downstream value of their recordings without spending hours on manual repurposing work.
Fireflies.ai
Fireflies.ai is a meeting intelligence platform that transcribes meetings automatically and uses AI to generate summaries, action items, and searchable notes across all recorded meetings.
Features: Automatic meeting joining via calendar integration, support for Zoom, Google Meet, Teams, Webex, and others, AI-generated summaries with action items and key decisions, speaker identification, full-text search across all meeting transcripts, conversation analytics showing talk time ratios, topic detection, and sentiment, integration with Slack, HubSpot, Salesforce, Notion, and 40+ other tools, task creation in Asana, Jira, and Trello from action items, clip creation for sharing specific meeting moments, team workspace for collaborative meeting notes.
Pricing: Free: unlimited meetings, 800 minutes storage, limited AI features. Pro $10/user/month: unlimited storage, AI summaries, search. Business $19/user/month: video recording, advanced analytics. Enterprise: custom.
Pros vs Descript: Meeting-native workflow with automatic joining is significantly more convenient than Descript's upload-based workflow for meeting transcription. The $10/month Pro plan is substantially cheaper than Descript for teams that need meeting notes. The 40+ integration library for pushing notes to other tools is extensive.
Cons vs Descript: No video editing. Designed exclusively for meetings, not for podcast production, documentary, or media workflows. The free plan's 800-minute storage limit becomes a constraint for teams with many meetings.
Best for: Remote teams, sales organizations, and product teams that want automatic transcription, summaries, and searchable archives of all their meetings without any manual workflow.
Grain
Grain is a meeting recording tool focused on creating shareable video highlights and connecting meeting insights to CRM records for sales coaching and customer research.
Features: Meeting recording and transcription for Zoom, Google Meet, and Teams, AI-generated meeting summary and highlight clips, one-click clip creation by highlighting transcript text, shareable clip pages with embedded video and transcript, HubSpot and Salesforce integration that attaches call recordings and highlights to CRM deals and contacts, coaching views for sales managers to review rep calls and add feedback, playlist creation for onboarding new reps with example calls, voice of customer reporting that aggregates themes across multiple customer conversations.
Pricing: Free: 5 meeting recordings. Starter $15/user/month: unlimited recordings, 30 clips. Business $40/user/month: AI features, CRM integration, coaching tools. Enterprise: custom.
Pros vs Descript: The clip creation and sharing workflow is faster and more shareable than Descript's for distributing specific meeting moments. CRM integration for attaching call recordings to deals is uniquely valuable for sales teams. The coaching workflow for sales managers is not available in Descript.
Cons vs Descript: No audio or video editing. Not designed for podcast production or media workflows. The value proposition is narrowly focused on meeting recordings for sales and customer research.
Best for: Sales teams that need to review and coach on recorded calls, customer success teams capturing voice of customer evidence, and product teams sharing user research highlights with stakeholders.
Adobe Premiere Pro
Adobe Premiere Pro, the industry-standard video editing application, added transcript-based editing with its Auto Transcribe feature, bringing a version of Descript's core workflow to the professional editing environment.
Features: Automatic transcription within Premiere with search and navigation, transcript-based sequence editing where you can find and select transcript text to navigate to that point in the timeline, automatic captions generation from the transcript, text-based editing in the Captions track that allows deleting transcript segments to cut the corresponding video, silence detection and removal, Remix tool for adjusting music to a new length, generative extend for extending clips, the full Premiere Pro editing toolset alongside the transcript features.
Pricing: Included in Adobe Creative Cloud ($54.99/month for All Apps, $22.99/month for Premiere single app).
Pros vs Descript: If you already pay for Adobe Creative Cloud, there is no additional cost. The full professional editing toolset alongside transcript features is a powerful combination for editors who need both. Industry-standard tool for professional video delivery workflows.
Cons vs Descript: The transcript-based editing in Premiere is less seamless than Descript's transcript-first interface -- the transcript is a navigation and caption tool alongside the traditional timeline rather than the primary editing surface. No podcast-specific features. No meeting transcription. The AI transcription accuracy varies by audio quality.
Best for: Professional video editors already in the Adobe ecosystem who want to add transcript-based editing to their existing Premiere Pro workflow without switching tools.
Whisper-Based Local Apps
Several user-friendly applications have been built on top of OpenAI's Whisper model to provide accurate local transcription without command-line knowledge.
MacWhisper (macOS): A native macOS app that uses Whisper models locally for accurate, fast transcription with no usage fees and no internet connection required. The free version supports the base model. A one-time purchase unlocks larger, more accurate models and additional features like translation and batch processing. For macOS users who want the best free transcription quality with a simple interface, MacWhisper is the practical recommendation.
Whisper Transcription (iOS): iPhone and iPad app that runs Whisper locally on device for transcription without sending audio to any server. Useful for transcribing recordings on the go.
Pricing: MacWhisper: free (base model), approximately $15-25 one-time for Pro features. Whisper Transcription iOS: free with in-app purchases.
Pros vs Descript: No monthly subscription -- one-time cost or free. No usage limits. Complete privacy -- audio never leaves your device. Accuracy comparable to paid services on clear audio.
Cons vs Descript: No editing tools, no content repurposing, no meeting integration. The output is a text file. Large model accuracy requires newer hardware for reasonable processing speed.
Best for: Privacy-conscious individuals, researchers, journalists working with sensitive source material, and anyone who wants high-accuracy transcription without an ongoing subscription fee.
Comparison Table
| Tool | Free Plan | Paid Plans | Best Feature | Biggest Limitation |
|---|---|---|---|---|
| Descript | Limited | $24/month | Transcript-based video editing | Learning curve, accuracy gaps |
| Otter.ai | 300 min/month | $17-30/user/month | Real-time meeting transcription | Not for media production |
| Whisper (local) | Free (self-hosted) | API $0.006/min | Zero cost, 97 languages | Technical setup required |
| Rev | No | $0.25-1.50/min | Human accuracy, HIPAA | Per-minute cost at scale |
| Sonix | No | $10/hr or $22/month | 35+ language translation | No editing, no meetings |
| Trint | No | $52-80/month | Newsroom workflow, research | Expensive, narrow audience |
| Castmagic | No | $39-99/month | Content repurposing output | No editing tools |
| Fireflies.ai | 800 min storage | $10-19/user/month | Meeting intelligence, integrations | Meetings only |
| Grain | 5 recordings | $15-40/user/month | Clip sharing, sales coaching | Meetings only |
| Adobe Premiere | No | $22.99+/month | Full professional editing | Less fluid than Descript |
| MacWhisper | Yes (base model) | ~$15-25 one-time | No subscription, full privacy | No editing tools |
Who Should Switch Away from Descript
Switch to Castmagic if your primary pain point is the time spent on post-production content creation -- show notes, social posts, newsletters -- rather than audio editing. The AI-generated outputs meaningfully reduce that time investment.
Switch to Fireflies.ai or Otter.ai if your main transcription use case is meeting notes and you are using Descript's meeting recording features primarily. Both are cheaper and better designed for the meeting workflow.
Switch to Whisper (via MacWhisper or similar) if you want the most accurate available AI transcription at no ongoing cost and you are comfortable with a tool that delivers a text file rather than an editing environment.
Switch to Rev if you produce content for legal, medical, journalistic, or broadcast contexts where 99%+ accuracy is required and the per-minute cost is proportionate to the stakes.
Switch to Sonix if you are working with non-English content at scale and need multi-language transcription and translation in a media-oriented workflow.
Who Should Stay with Descript
Descript is the right tool when you want to edit audio and video by editing the transcript, and you are willing to invest in learning the paradigm. For interview-based podcast and video content where the edit points are driven by what was said rather than technical audio considerations, the transcript-editing workflow is genuinely faster once internalized. If you have already invested in learning Descript and the accuracy is sufficient for your content, there is no compelling reason to switch tools just to save $24/month.
Related reading: Best Alternatives to Riverside.fm for Remote Recording in 2026 | Best Podcast Tools in 2026 | Best AI Tools for Creators in 2026
Frequently Asked Questions
What is the best free alternative to Descript for transcription?
OpenAI's Whisper model is the most capable free transcription option available. As an open-source model, Whisper can be run locally on your own machine at no cost, with no API fees, no usage limits, and no data being sent to a third party. The trade-off is setup complexity: running Whisper locally requires comfort with command-line tools or installing a third-party app built on Whisper. MacWhisper on macOS and Whisper Transcription on iOS provide friendly interfaces over the Whisper model with free tiers. For users who want free transcription without any technical setup, Otter.ai's free plan includes 300 transcription minutes per month and real-time transcription during meetings, which is functional for light use. Fireflies.ai's free plan transcribes unlimited meetings but stores only 800 minutes of transcripts, which limits the archive. The honest assessment is that truly free and high-quality transcription without technical overhead is a difficult combination -- the best free options are either technically demanding (Whisper), limited in minutes (Otter.ai), or limited in storage (Fireflies.ai).
What tools transcribe meetings and video automatically?
Fireflies.ai and Otter.ai are the two most widely used tools for automatic meeting transcription. Fireflies.ai joins meetings as a bot participant on Zoom, Google Meet, Microsoft Teams, and Webex, transcribes in real time, generates AI summaries with action items, and pushes the notes to Slack, HubSpot, Salesforce, and other integrations. The free plan covers unlimited meetings with 800 minutes of storage. Paid plans start at $10/month per user. Otter.ai integrates with Zoom, Google Calendar, and Microsoft Calendar to automatically detect and join scheduled meetings. It produces real-time transcripts with speaker identification, highlighted action items, and shared notebooks for collaborative annotation. Grain focuses on meeting highlights -- it records meetings, transcribes them, and allows users to clip specific moments as short video highlights that can be shared with team members or embedded in documents. This makes it particularly valuable for sales call review and coaching. For video content rather than meetings, Sonix and Trint are designed for media workflows where the transcript needs to be edited, searched, and exported in broadcast-ready formats.
What alternatives to Descript are best for podcasters?
Castmagic is purpose-built for podcasters who want to repurpose audio and video content into written assets. It generates a transcript, then uses AI to create show notes, social media posts, email newsletters, blog article drafts, chapter markers, and guest quotes from the same recording. The value is not transcription accuracy alone but the downstream content generation that turns one recording session into a week of content. At $39-99/month depending on plan, it replaces both a transcription service and a significant amount of manual repurposing work. Descript itself remains the strongest option for podcasters who want to edit the audio and video by editing the transcript -- deleting a word in the transcript deletes it from the recording, and removing filler words can be done in bulk. If the transcript-based editing workflow is the reason you are considering Descript, no alternative fully replicates it. Podcastle is an AI-powered tool that handles recording, basic editing, noise removal, and transcription in one interface, targeting podcasters who want a simpler all-in-one solution than Descript. Alitu takes the simplification further with a guided workflow that handles audio cleaning, episode assembly, and distribution with minimal technical knowledge required.
What is the most accurate AI transcription tool?
Rev's human transcription service at $1.50/minute is the most accurate option for content where accuracy is critical and budget allows. Human transcribers achieve accuracy rates above 99% even for content with strong accents, technical vocabulary, crosstalk, and background noise that degrade AI model accuracy. For AI-only transcription, the accuracy landscape has converged significantly. OpenAI's Whisper Large model, Rev's AI, Sonix, and Trint all achieve accuracy rates in the 85-95% range on clear audio with standard accents and general vocabulary. Accuracy degrades on content with multiple speakers, heavy accents, domain-specific terminology (medical, legal, technical), poor audio quality, and significant background noise. For podcasters and video creators working with professional microphones and controlled audio environments, most AI transcription tools are sufficiently accurate. For legal depositions, medical dictation, academic research interviews, or broadcast media requiring verbatim accuracy, human transcription from Rev or similar services remains the reliable choice. Whisper Large consistently performs at the top of benchmark comparisons for AI transcription, and its availability as a free open-source model makes it the default recommendation when accuracy matters and budget is constrained.
What Descript alternatives are best for editing video via transcript?
Descript's transcript-based video editing -- where editing the text transcript edits the underlying audio and video -- is Descript's most distinctive feature. No other tool fully replicates it. Adobe Premiere Pro added transcript-based editing with its Auto Transcribe feature, allowing editors to find and delete sections by searching the transcript, but the integration is not as seamless as Descript's where the transcript is the primary editing interface rather than a navigation aid alongside the traditional timeline. Kapwing and Clideo offer basic transcript-based editing for online content creators, though at lower sophistication than Descript. The honest assessment is that if transcript-based editing is your primary requirement, Descript remains the best tool for that specific workflow. The alternatives that compete with Descript do so on specific sub-tasks -- Castmagic for content repurposing, Otter.ai for meeting transcription, Sonix for media transcription accuracy and export formats -- rather than on the integrated editing workflow that Descript provides.
What tools help repurpose long video content into clips and social posts?
Castmagic is the most complete tool for converting long-form audio and video content into multiple written and social formats. A single upload generates a transcript, show notes, social media captions, newsletter content, blog outlines, and quote graphics from the same source material. Opus Clip (not in this guide but worth mentioning) focuses specifically on identifying and clipping the best moments from long videos into short-form clips for social media with AI-generated captions. Grain generates shareable video highlights from meeting recordings with one click, which is valuable for sales teams sharing call moments and for product teams capturing customer feedback snippets. Descript itself supports this workflow with its Underlord AI feature that can identify highlight moments and create clips, though the workflow requires more manual involvement than Castmagic's automated output generation.
What transcription tools support the most languages?
Sonix supports transcription in 35+ languages and translation into 35+ languages, making it one of the strongest options for international media workflows. The translation feature converts transcripts from one language to another within the same editor, which is valuable for organizations working with multilingual content. Whisper (OpenAI's open-source model) supports 97 languages for transcription and translation, which is the broadest language coverage of any transcription model available. The catch is that accuracy varies significantly by language -- Whisper's accuracy on English, Spanish, French, and German is significantly higher than on lower-resource languages. For well-resourced languages, Whisper's free open-source model is competitive with paid services. Fireflies.ai supports over 60 languages for meeting transcription, which covers most global business use cases. Otter.ai is English-first and less suited for non-English content despite recent additions.